WEBVTT Kind: captions Language: en 00:00:09.140 --> 00:00:13.760 align:start position:0% [Music] 00:00:13.760 --> 00:00:13.770 align:start position:0% 00:00:13.770 --> 00:00:16.710 align:start position:0% hello<00:00:14.770> everybody<00:00:15.100> I'm<00:00:15.760> Alexa<00:00:16.119> grab<00:00:16.330> roof<00:00:16.510> I'm 00:00:16.710 --> 00:00:16.720 align:start position:0% hello everybody I'm Alexa grab roof I'm 00:00:16.720 --> 00:00:19.260 align:start position:0% hello everybody I'm Alexa grab roof I'm the<00:00:17.050> organizer<00:00:17.590> of<00:00:17.830> barrera<00:00:18.310> a<00:00:18.430> I'm<00:00:18.699> atop<00:00:18.970> and 00:00:19.260 --> 00:00:19.270 align:start position:0% the organizer of barrera a I'm atop and 00:00:19.270 --> 00:00:22.050 align:start position:0% the organizer of barrera a I'm atop and here<00:00:19.960> we're<00:00:20.140> at<00:00:20.290> night<00:00:20.500> row<00:00:20.830> which<00:00:21.610> is<00:00:21.730> hosting 00:00:22.050 --> 00:00:22.060 align:start position:0% here we're at night row which is hosting 00:00:22.060 --> 00:00:24.660 align:start position:0% here we're at night row which is hosting us<00:00:22.210> for<00:00:22.720> very<00:00:22.750> exciting<00:00:23.320> the<00:00:23.410> table<00:00:23.670> with 00:00:24.660 --> 00:00:24.670 align:start position:0% us for very exciting the table with 00:00:24.670 --> 00:00:27.540 align:start position:0% us for very exciting the table with determined<00:00:25.180> AI<00:00:25.440> which<00:00:26.440> is<00:00:26.560> a<00:00:26.590> new<00:00:26.800> startup<00:00:27.250> and 00:00:27.540 --> 00:00:27.550 align:start position:0% determined AI which is a new startup and 00:00:27.550 --> 00:00:30.270 align:start position:0% determined AI which is a new startup and Evan<00:00:28.030> sparks<00:00:28.390> the<00:00:28.539> CEO<00:00:28.750> is<00:00:28.960> here<00:00:28.990> with<00:00:29.289> us<00:00:29.410> he's 00:00:30.270 --> 00:00:30.280 align:start position:0% Evan sparks the CEO is here with us he's 00:00:30.280 --> 00:00:32.220 align:start position:0% Evan sparks the CEO is here with us he's gonna<00:00:30.460> give<00:00:31.000> us<00:00:31.150> a<00:00:31.180> talk<00:00:31.449> about<00:00:31.630> this<00:00:31.990> later 00:00:32.220 --> 00:00:32.230 align:start position:0% gonna give us a talk about this later 00:00:32.230 --> 00:00:35.010 align:start position:0% gonna give us a talk about this later which<00:00:33.129> you<00:00:33.219> can<00:00:33.370> also<00:00:33.489> find<00:00:33.820> on<00:00:34.180> functional<00:00:34.690> TV 00:00:35.010 --> 00:00:35.020 align:start position:0% which you can also find on functional TV 00:00:35.020 --> 00:00:37.610 align:start position:0% which you can also find on functional TV where<00:00:35.230> everything<00:00:35.710> goes<00:00:35.859> phone<00:00:36.760> call<00:00:36.940> dot<00:00:37.120> TV 00:00:37.610 --> 00:00:37.620 align:start position:0% where everything goes phone call dot TV 00:00:37.620 --> 00:00:40.529 align:start position:0% where everything goes phone call dot TV but<00:00:38.620> first<00:00:39.370> of<00:00:39.489> all<00:00:39.550> welcome<00:00:39.730> Evan<00:00:40.059> thank<00:00:40.420> you 00:00:40.529 --> 00:00:40.539 align:start position:0% but first of all welcome Evan thank you 00:00:40.539 --> 00:00:44.009 align:start position:0% but first of all welcome Evan thank you thank<00:00:40.839> you<00:00:40.899> thank<00:00:41.170> you<00:00:41.199> very<00:00:41.260> much<00:00:42.719> yeah<00:00:43.719> so 00:00:44.009 --> 00:00:44.019 align:start position:0% thank you thank you very much yeah so 00:00:44.019 --> 00:00:46.520 align:start position:0% thank you thank you very much yeah so determined<00:00:44.800> AI<00:00:44.859> we<00:00:45.219> are<00:00:45.460> building<00:00:46.030> a 00:00:46.520 --> 00:00:46.530 align:start position:0% determined AI we are building a 00:00:46.530 --> 00:00:49.619 align:start position:0% determined AI we are building a workbench<00:00:47.530> for<00:00:47.949> machine<00:00:48.940> learning<00:00:48.999> engineers 00:00:49.619 --> 00:00:49.629 align:start position:0% workbench for machine learning engineers 00:00:49.629 --> 00:00:52.290 align:start position:0% workbench for machine learning engineers and<00:00:50.079> data<00:00:50.589> scientists<00:00:51.249> who<00:00:51.429> are<00:00:51.670> trying<00:00:51.940> to 00:00:52.290 --> 00:00:52.300 align:start position:0% and data scientists who are trying to 00:00:52.300 --> 00:00:56.039 align:start position:0% and data scientists who are trying to get<00:00:52.989> their<00:00:53.579> AI<00:00:54.579> powered<00:00:55.149> features<00:00:55.809> and 00:00:56.039 --> 00:00:56.049 align:start position:0% get their AI powered features and 00:00:56.049 --> 00:00:58.169 align:start position:0% get their AI powered features and applications<00:00:56.229> out<00:00:56.799> to<00:00:57.100> market<00:00:57.429> quicker<00:00:57.789> than 00:00:58.169 --> 00:00:58.179 align:start position:0% applications out to market quicker than 00:00:58.179 --> 00:01:00.149 align:start position:0% applications out to market quicker than they<00:00:58.329> could<00:00:58.479> before<00:00:58.600> so<00:00:59.079> if<00:00:59.350> you've<00:00:59.710> got<00:00:59.889> tons 00:01:00.149 --> 00:01:00.159 align:start position:0% they could before so if you've got tons 00:01:00.159 --> 00:01:02.610 align:start position:0% they could before so if you've got tons of<00:01:00.339> data<00:01:00.489> lots<00:01:00.940> of<00:01:01.030> GPUs<00:01:01.449> deep<00:01:02.350> learning 00:01:02.610 --> 00:01:02.620 align:start position:0% of data lots of GPUs deep learning 00:01:02.620 --> 00:01:05.280 align:start position:0% of data lots of GPUs deep learning models<00:01:02.920> that<00:01:03.219> you<00:01:03.489> want<00:01:03.760> to<00:01:04.049> encode<00:01:05.049> it<00:01:05.199> to 00:01:05.280 --> 00:01:05.290 align:start position:0% models that you want to encode it to 00:01:05.290 --> 00:01:08.179 align:start position:0% models that you want to encode it to your<00:01:05.409> favorite<00:01:06.040> image<00:01:06.100> classification<00:01:06.430> or<00:01:07.210> or 00:01:08.179 --> 00:01:08.189 align:start position:0% your favorite image classification or or 00:01:08.189 --> 00:01:10.350 align:start position:0% your favorite image classification or or text<00:01:09.189> of<00:01:09.250> speech<00:01:09.430> kind<00:01:09.610> of<00:01:09.700> applications<00:01:10.180> and 00:01:10.350 --> 00:01:10.360 align:start position:0% text of speech kind of applications and 00:01:10.360 --> 00:01:12.779 align:start position:0% text of speech kind of applications and so<00:01:10.719> on<00:01:10.750> that's<00:01:11.380> we<00:01:11.799> help<00:01:11.950> you<00:01:12.039> get<00:01:12.189> those 00:01:12.779 --> 00:01:12.789 align:start position:0% so on that's we help you get those 00:01:12.789 --> 00:01:13.980 align:start position:0% so on that's we help you get those things<00:01:12.939> built<00:01:13.240> faster<00:01:13.630> than<00:01:13.780> you<00:01:13.840> could 00:01:13.980 --> 00:01:13.990 align:start position:0% things built faster than you could 00:01:13.990 --> 00:01:16.920 align:start position:0% things built faster than you could before<00:01:14.110> so<00:01:14.890> what's<00:01:15.130> our<00:01:15.549> engineer<00:01:16.030> first<00:01:16.689> and 00:01:16.920 --> 00:01:16.930 align:start position:0% before so what's our engineer first and 00:01:16.930 --> 00:01:19.289 align:start position:0% before so what's our engineer first and then<00:01:17.049> a<00:01:17.079> scientist<00:01:17.590> later<00:01:17.829> right<00:01:18.490> so<00:01:18.789> explain 00:01:19.289 --> 00:01:19.299 align:start position:0% then a scientist later right so explain 00:01:19.299 --> 00:01:22.200 align:start position:0% then a scientist later right so explain it<00:01:19.509> to<00:01:19.689> me<00:01:19.810> when<00:01:20.530> I<00:01:20.859> see<00:01:21.280> the<00:01:21.640> result<00:01:21.909> of<00:01:22.030> folks 00:01:22.200 --> 00:01:22.210 align:start position:0% it to me when I see the result of folks 00:01:22.210 --> 00:01:25.889 align:start position:0% it to me when I see the result of folks who<00:01:22.869> say<00:01:23.049> that<00:01:23.289> they're<00:01:23.469> deploying<00:01:23.939> models<00:01:24.939> to 00:01:25.889 --> 00:01:25.899 align:start position:0% who say that they're deploying models to 00:01:25.899 --> 00:01:27.209 align:start position:0% who say that they're deploying models to production<00:01:26.200> and<00:01:26.680> there<00:01:26.859> is<00:01:26.920> this<00:01:27.039> whole 00:01:27.209 --> 00:01:27.219 align:start position:0% production and there is this whole 00:01:27.219 --> 00:01:30.749 align:start position:0% production and there is this whole industry<00:01:27.430> emerging<00:01:28.350> why<00:01:29.350> why<00:01:29.950> do<00:01:30.399> we<00:01:30.460> see<00:01:30.609> this 00:01:30.749 --> 00:01:30.759 align:start position:0% industry emerging why why do we see this 00:01:30.759 --> 00:01:32.069 align:start position:0% industry emerging why why do we see this phenomenon<00:01:31.240> why<00:01:31.390> there<00:01:31.630> is<00:01:31.719> basically<00:01:31.990> a 00:01:32.069 --> 00:01:32.079 align:start position:0% phenomenon why there is basically a 00:01:32.079 --> 00:01:33.779 align:start position:0% phenomenon why there is basically a whole<00:01:32.109> class<00:01:32.530> of<00:01:32.799> companies<00:01:33.009> which<00:01:33.369> say<00:01:33.579> we'll 00:01:33.779 --> 00:01:33.789 align:start position:0% whole class of companies which say we'll 00:01:33.789 --> 00:01:35.669 align:start position:0% whole class of companies which say we'll help<00:01:34.000> you<00:01:34.179> perfect<00:01:34.390> answer<00:01:34.719> I<00:01:34.750> was<00:01:34.960> just<00:01:35.469> not 00:01:35.669 --> 00:01:35.679 align:start position:0% help you perfect answer I was just not 00:01:35.679 --> 00:01:37.590 align:start position:0% help you perfect answer I was just not another<00:01:35.979> kind<00:01:36.219> of<00:01:36.310> software<00:01:36.670> that<00:01:37.000> you<00:01:37.479> know<00:01:37.570> a 00:01:37.590 --> 00:01:37.600 align:start position:0% another kind of software that you know a 00:01:37.600 --> 00:01:39.179 align:start position:0% another kind of software that you know a lot<00:01:37.689> of<00:01:37.750> people<00:01:38.049> know<00:01:38.170> how<00:01:38.320> to<00:01:38.380> deal<00:01:38.649> with<00:01:38.679> yeah 00:01:39.179 --> 00:01:39.189 align:start position:0% lot of people know how to deal with yeah 00:01:39.189 --> 00:01:40.830 align:start position:0% lot of people know how to deal with yeah so<00:01:39.520> I<00:01:39.549> think<00:01:39.670> that<00:01:39.939> what<00:01:40.240> we're<00:01:40.450> seeing<00:01:40.659> is 00:01:40.830 --> 00:01:40.840 align:start position:0% so I think that what we're seeing is 00:01:40.840 --> 00:01:42.389 align:start position:0% so I think that what we're seeing is that<00:01:40.890> fundamentally<00:01:41.890> the<00:01:42.039> way<00:01:42.189> we're 00:01:42.389 --> 00:01:42.399 align:start position:0% that fundamentally the way we're 00:01:42.399 --> 00:01:44.580 align:start position:0% that fundamentally the way we're developing<00:01:42.880> software<00:01:43.149> today<00:01:43.659> is<00:01:43.869> changing<00:01:44.409> as 00:01:44.580 --> 00:01:44.590 align:start position:0% developing software today is changing as 00:01:44.590 --> 00:01:48.679 align:start position:0% developing software today is changing as a<00:01:44.619> as<00:01:45.070> a<00:01:45.100> result<00:01:45.369> of<00:01:46.079> doing<00:01:47.079> data<00:01:47.950> driven 00:01:48.679 --> 00:01:48.689 align:start position:0% a as a result of doing data driven 00:01:48.689 --> 00:01:51.719 align:start position:0% a as a result of doing data driven application<00:01:49.689> development<00:01:50.399> things<00:01:51.399> change 00:01:51.719 --> 00:01:51.729 align:start position:0% application development things change 00:01:51.729 --> 00:01:53.489 align:start position:0% application development things change you<00:01:51.880> know<00:01:51.969> we've<00:01:52.210> got<00:01:52.390> a<00:01:52.420> lot<00:01:52.719> of<00:01:52.749> uncertainty 00:01:53.489 --> 00:01:53.499 align:start position:0% you know we've got a lot of uncertainty 00:01:53.499 --> 00:01:54.929 align:start position:0% you know we've got a lot of uncertainty we've<00:01:53.679> got<00:01:53.829> statistical<00:01:54.340> randomness<00:01:54.820> we've 00:01:54.929 --> 00:01:54.939 align:start position:0% we've got statistical randomness we've 00:01:54.939 --> 00:01:57.029 align:start position:0% we've got statistical randomness we've got<00:01:55.060> noise<00:01:55.390> things<00:01:56.109> that<00:01:56.289> you<00:01:56.380> have<00:01:56.560> to<00:01:56.679> deal 00:01:57.029 --> 00:01:57.039 align:start position:0% got noise things that you have to deal 00:01:57.039 --> 00:01:59.219 align:start position:0% got noise things that you have to deal with<00:01:57.070> that<00:01:57.429> are<00:01:58.179> quite<00:01:58.840> different<00:01:59.049> than 00:01:59.219 --> 00:01:59.229 align:start position:0% with that are quite different than 00:01:59.229 --> 00:02:00.510 align:start position:0% with that are quite different than writing<00:01:59.469> down<00:01:59.710> a<00:01:59.770> bunch<00:02:00.009> of<00:02:00.069> rules<00:02:00.310> and 00:02:00.510 --> 00:02:00.520 align:start position:0% writing down a bunch of rules and 00:02:00.520 --> 00:02:02.910 align:start position:0% writing down a bunch of rules and software<00:02:00.909> and<00:02:01.119> so<00:02:01.270> we<00:02:01.780> now<00:02:02.289> have<00:02:02.799> the 00:02:02.910 --> 00:02:02.920 align:start position:0% software and so we now have the 00:02:02.920 --> 00:02:04.620 align:start position:0% software and so we now have the opportunity<00:02:03.460> to<00:02:03.640> make<00:02:03.789> better<00:02:03.999> decisions<00:02:04.509> on 00:02:04.620 --> 00:02:04.630 align:start position:0% opportunity to make better decisions on 00:02:04.630 --> 00:02:06.510 align:start position:0% opportunity to make better decisions on the<00:02:04.719> basis<00:02:05.020> of<00:02:05.140> data<00:02:05.289> build<00:02:05.740> models<00:02:06.100> out<00:02:06.249> that 00:02:06.510 --> 00:02:06.520 align:start position:0% the basis of data build models out that 00:02:06.520 --> 00:02:08.340 align:start position:0% the basis of data build models out that that<00:02:06.969> can<00:02:07.119> make<00:02:07.149> more<00:02:07.479> accurate<00:02:07.600> predictions 00:02:08.340 --> 00:02:08.350 align:start position:0% that can make more accurate predictions 00:02:08.350 --> 00:02:10.800 align:start position:0% that can make more accurate predictions and<00:02:08.830> so<00:02:08.979> on<00:02:09.130> but<00:02:09.580> it<00:02:09.700> means<00:02:09.819> that<00:02:10.119> almost<00:02:10.390> every 00:02:10.800 --> 00:02:10.810 align:start position:0% and so on but it means that almost every 00:02:10.810 --> 00:02:12.930 align:start position:0% and so on but it means that almost every stage<00:02:11.140> of<00:02:11.500> the<00:02:11.950> model<00:02:12.190> development<00:02:12.520> process 00:02:12.930 --> 00:02:12.940 align:start position:0% stage of the model development process 00:02:12.940 --> 00:02:15.050 align:start position:0% stage of the model development process it<00:02:13.360> and<00:02:13.720> software<00:02:14.170> Duvall 00:02:15.050 --> 00:02:15.060 align:start position:0% it and software Duvall 00:02:15.060 --> 00:02:16.820 align:start position:0% it and software Duvall process<00:02:15.420> is<00:02:15.540> going<00:02:15.690> to<00:02:15.810> change<00:02:16.080> so<00:02:16.230> everything 00:02:16.820 --> 00:02:16.830 align:start position:0% process is going to change so everything 00:02:16.830 --> 00:02:20.390 align:start position:0% process is going to change so everything from<00:02:17.069> versioning<00:02:17.940> now<00:02:18.630> comes<00:02:18.900> down<00:02:19.739> to<00:02:19.950> more 00:02:20.390 --> 00:02:20.400 align:start position:0% from versioning now comes down to more 00:02:20.400 --> 00:02:21.830 align:start position:0% from versioning now comes down to more than<00:02:20.580> just<00:02:20.640> software<00:02:21.060> it<00:02:21.239> also<00:02:21.360> comes<00:02:21.599> out<00:02:21.750> to 00:02:21.830 --> 00:02:21.840 align:start position:0% than just software it also comes out to 00:02:21.840 --> 00:02:25.580 align:start position:0% than just software it also comes out to the<00:02:21.930> data<00:02:22.140> everything<00:02:23.040> around<00:02:23.910> see<00:02:24.840> ICD<00:02:25.230> and 00:02:25.580 --> 00:02:25.590 align:start position:0% the data everything around see ICD and 00:02:25.590 --> 00:02:27.530 align:start position:0% the data everything around see ICD and deployment<00:02:26.069> and<00:02:26.340> monitoring<00:02:26.730> also<00:02:27.150> changes 00:02:27.530 --> 00:02:27.540 align:start position:0% deployment and monitoring also changes 00:02:27.540 --> 00:02:29.330 align:start position:0% deployment and monitoring also changes because<00:02:27.569> the<00:02:28.200> world<00:02:28.349> changes<00:02:28.739> around<00:02:28.770> us<00:02:29.160> our 00:02:29.330 --> 00:02:29.340 align:start position:0% because the world changes around us our 00:02:29.340 --> 00:02:31.010 align:start position:0% because the world changes around us our models<00:02:29.610> react<00:02:29.910> and<00:02:30.120> they're<00:02:30.239> more<00:02:30.630> difficult 00:02:31.010 --> 00:02:31.020 align:start position:0% models react and they're more difficult 00:02:31.020 --> 00:02:32.660 align:start position:0% models react and they're more difficult to<00:02:31.140> debug<00:02:31.349> and<00:02:31.650> so<00:02:31.860> on<00:02:32.010> than<00:02:32.160> traditional 00:02:32.660 --> 00:02:32.670 align:start position:0% to debug and so on than traditional 00:02:32.670 --> 00:02:34.640 align:start position:0% to debug and so on than traditional software<00:02:32.970> so<00:02:33.510> every<00:02:33.810> stage<00:02:34.080> of<00:02:34.230> that<00:02:34.380> pipeline 00:02:34.640 --> 00:02:34.650 align:start position:0% software so every stage of that pipeline 00:02:34.650 --> 00:02:37.040 align:start position:0% software so every stage of that pipeline has<00:02:35.040> to<00:02:35.190> be<00:02:35.310> rethought<00:02:35.700> thanks<00:02:36.150> to<00:02:36.330> thanks<00:02:36.870> to 00:02:37.040 --> 00:02:37.050 align:start position:0% has to be rethought thanks to thanks to 00:02:37.050 --> 00:02:38.540 align:start position:0% has to be rethought thanks to thanks to these<00:02:37.230> new<00:02:37.410> methods<00:02:37.769> that<00:02:37.890> are<00:02:38.280> powering 00:02:38.540 --> 00:02:38.550 align:start position:0% these new methods that are powering 00:02:38.550 --> 00:02:40.940 align:start position:0% these new methods that are powering better<00:02:38.819> applications<00:02:39.510> mm-hmm<00:02:40.019> yes<00:02:40.709> you<00:02:40.890> know 00:02:40.940 --> 00:02:40.950 align:start position:0% better applications mm-hmm yes you know 00:02:40.950 --> 00:02:44.630 align:start position:0% better applications mm-hmm yes you know I<00:02:41.400> think<00:02:42.380> here<00:02:43.380> I<00:02:43.590> was<00:02:43.769> pushing<00:02:44.069> this<00:02:44.310> a<00:02:44.370> little 00:02:44.630 --> 00:02:44.640 align:start position:0% I think here I was pushing this a little 00:02:44.640 --> 00:02:46.580 align:start position:0% I think here I was pushing this a little bit<00:02:44.790> too<00:02:45.030> to<00:02:45.269> the<00:02:45.360> data<00:02:45.540> scientist<00:02:45.930> so<00:02:46.110> so 00:02:46.580 --> 00:02:46.590 align:start position:0% bit too to the data scientist so so 00:02:46.590 --> 00:02:49.250 align:start position:0% bit too to the data scientist so so here's<00:02:46.769> my<00:02:46.920> kind<00:02:47.100> of<00:02:47.190> technology<00:02:47.940> so<00:02:48.090> the<00:02:48.330> the 00:02:49.250 --> 00:02:49.260 align:start position:0% here's my kind of technology so the the 00:02:49.260 --> 00:02:51.020 align:start position:0% here's my kind of technology so the the result<00:02:49.560> of<00:02:49.680> new<00:02:49.799> folks<00:02:50.010> who<00:02:50.130> came<00:02:50.310> into<00:02:50.580> the 00:02:51.020 --> 00:02:51.030 align:start position:0% result of new folks who came into the 00:02:51.030 --> 00:02:55.280 align:start position:0% result of new folks who came into the industry<00:02:51.720> and<00:02:52.410> they<00:02:53.220> learn<00:02:53.940> both<00:02:54.780> the<00:02:54.959> designs 00:02:55.280 --> 00:02:55.290 align:start position:0% industry and they learn both the designs 00:02:55.290 --> 00:02:58.520 align:start position:0% industry and they learn both the designs and<00:02:55.620> programming<00:02:56.519> in<00:02:56.730> Python<00:02:56.880> and<00:02:57.750> they<00:02:58.470> do 00:02:58.520 --> 00:02:58.530 align:start position:0% and programming in Python and they do 00:02:58.530 --> 00:03:00.680 align:start position:0% and programming in Python and they do not<00:02:58.739> go<00:02:58.860> through<00:02:58.980> the<00:02:59.160> traditional<00:02:59.849> career 00:03:00.680 --> 00:03:00.690 align:start position:0% not go through the traditional career 00:03:00.690 --> 00:03:02.750 align:start position:0% not go through the traditional career arc<00:03:00.900> of<00:03:01.110> a<00:03:01.170> software<00:03:01.440> engineer<00:03:01.769> we<00:03:01.980> learn<00:03:02.370> the 00:03:02.750 --> 00:03:02.760 align:start position:0% arc of a software engineer we learn the 00:03:02.760 --> 00:03:05.030 align:start position:0% arc of a software engineer we learn the ICD<00:03:03.239> and<00:03:03.569> test-driven<00:03:04.200> development<00:03:04.470> and<00:03:04.980> all 00:03:05.030 --> 00:03:05.040 align:start position:0% ICD and test-driven development and all 00:03:05.040 --> 00:03:06.800 align:start position:0% ICD and test-driven development and all this<00:03:05.130> kind<00:03:05.340> of<00:03:05.370> stuff<00:03:05.580> right<00:03:05.700> so<00:03:05.819> you<00:03:06.569> know<00:03:06.660> in 00:03:06.800 --> 00:03:06.810 align:start position:0% this kind of stuff right so you know in 00:03:06.810 --> 00:03:08.199 align:start position:0% this kind of stuff right so you know in my<00:03:06.930> mind<00:03:06.959> that's<00:03:07.650> my<00:03:07.769> kind<00:03:07.950> of<00:03:07.980> first 00:03:08.199 --> 00:03:08.209 align:start position:0% my mind that's my kind of first 00:03:08.209 --> 00:03:10.789 align:start position:0% my mind that's my kind of first explanation<00:03:09.209> to<00:03:09.480> myself<00:03:09.630> that<00:03:10.110> this<00:03:10.290> industry 00:03:10.789 --> 00:03:10.799 align:start position:0% explanation to myself that this industry 00:03:10.799 --> 00:03:12.680 align:start position:0% explanation to myself that this industry arises<00:03:11.640> because<00:03:11.849> there's<00:03:12.299> whole<00:03:12.450> bunch<00:03:12.630> of 00:03:12.680 --> 00:03:12.690 align:start position:0% arises because there's whole bunch of 00:03:12.690 --> 00:03:15.140 align:start position:0% arises because there's whole bunch of folks<00:03:12.930> who<00:03:13.200> are<00:03:13.230> either<00:03:13.830> data<00:03:14.549> sciences<00:03:15.000> with 00:03:15.140 --> 00:03:15.150 align:start position:0% folks who are either data sciences with 00:03:15.150 --> 00:03:17.300 align:start position:0% folks who are either data sciences with also<00:03:15.450> 3d<00:03:15.660> nearing<00:03:16.019> experience<00:03:16.500> all<00:03:16.890> that 00:03:17.300 --> 00:03:17.310 align:start position:0% also 3d nearing experience all that 00:03:17.310 --> 00:03:19.670 align:start position:0% also 3d nearing experience all that business<00:03:17.489> folks<00:03:17.819> so<00:03:18.000> the<00:03:18.150> model<00:03:18.569> is<00:03:18.810> serving 00:03:19.670 --> 00:03:19.680 align:start position:0% business folks so the model is serving 00:03:19.680 --> 00:03:21.440 align:start position:0% business folks so the model is serving business<00:03:20.100> so<00:03:20.340> business<00:03:20.760> people<00:03:20.940> are<00:03:21.120> watching 00:03:21.440 --> 00:03:21.450 align:start position:0% business so business people are watching 00:03:21.450 --> 00:03:23.569 align:start position:0% business so business people are watching it<00:03:21.570> right<00:03:22.440> the<00:03:22.620> data<00:03:22.769> Sciences<00:03:23.190> are<00:03:23.310> working 00:03:23.569 --> 00:03:23.579 align:start position:0% it right the data Sciences are working 00:03:23.579 --> 00:03:26.979 align:start position:0% it right the data Sciences are working on<00:03:23.700> it<00:03:24.440> but<00:03:25.440> they<00:03:25.560> don't<00:03:25.739> have<00:03:25.980> necessarily 00:03:26.979 --> 00:03:26.989 align:start position:0% on it but they don't have necessarily 00:03:26.989 --> 00:03:29.479 align:start position:0% on it but they don't have necessarily rigorous<00:03:27.989> experience<00:03:28.500> of<00:03:28.709> deploying<00:03:29.100> and 00:03:29.479 --> 00:03:29.489 align:start position:0% rigorous experience of deploying and 00:03:29.489 --> 00:03:31.970 align:start position:0% rigorous experience of deploying and monitoring<00:03:29.910> roll<00:03:30.239> it<00:03:30.359> back<00:03:30.510> and<00:03:30.720> so<00:03:31.260> kind<00:03:31.920> of 00:03:31.970 --> 00:03:31.980 align:start position:0% monitoring roll it back and so kind of 00:03:31.980 --> 00:03:33.860 align:start position:0% monitoring roll it back and so kind of models<00:03:32.400> intersecting<00:03:32.970> so<00:03:33.180> engineer<00:03:33.540> data 00:03:33.860 --> 00:03:33.870 align:start position:0% models intersecting so engineer data 00:03:33.870 --> 00:03:35.240 align:start position:0% models intersecting so engineer data science<00:03:34.170> and<00:03:34.290> business<00:03:34.380> people<00:03:34.590> is<00:03:35.040> it 00:03:35.240 --> 00:03:35.250 align:start position:0% science and business people is it 00:03:35.250 --> 00:03:35.960 align:start position:0% science and business people is it accurate 00:03:35.960 --> 00:03:35.970 align:start position:0% accurate 00:03:35.970 --> 00:03:37.550 align:start position:0% accurate you<00:03:36.060> might<00:03:36.359> or<00:03:36.690> yeah<00:03:36.989> I<00:03:37.019> think<00:03:37.260> there's 00:03:37.550 --> 00:03:37.560 align:start position:0% you might or yeah I think there's 00:03:37.560 --> 00:03:39.620 align:start position:0% you might or yeah I think there's there's<00:03:37.950> one<00:03:38.340> way<00:03:38.489> of<00:03:38.640> looking<00:03:39.209> at<00:03:39.329> it<00:03:39.450> where 00:03:39.620 --> 00:03:39.630 align:start position:0% there's one way of looking at it where 00:03:39.630 --> 00:03:42.380 align:start position:0% there's one way of looking at it where it<00:03:39.780> it<00:03:39.989> says<00:03:40.319> let's<00:03:41.130> take<00:03:41.640> you<00:03:42.060> know<00:03:42.180> these 00:03:42.380 --> 00:03:42.390 align:start position:0% it it says let's take you know these 00:03:42.390 --> 00:03:44.660 align:start position:0% it it says let's take you know these data<00:03:42.569> scientists<00:03:43.170> who<00:03:43.319> are<00:03:43.620> running<00:03:44.549> wild 00:03:44.660 --> 00:03:44.670 align:start position:0% data scientists who are running wild 00:03:44.670 --> 00:03:46.129 align:start position:0% data scientists who are running wild with<00:03:44.940> there<00:03:45.150> are<00:03:45.269> scripts<00:03:45.540> and<00:03:45.720> there<00:03:45.840> Python 00:03:46.129 --> 00:03:46.139 align:start position:0% with there are scripts and there Python 00:03:46.139 --> 00:03:48.170 align:start position:0% with there are scripts and there Python scripts<00:03:46.380> and<00:03:46.590> so<00:03:46.709> on<00:03:46.859> and<00:03:47.130> don't<00:03:48.090> know 00:03:48.170 --> 00:03:48.180 align:start position:0% scripts and so on and don't know 00:03:48.180 --> 00:03:49.400 align:start position:0% scripts and so on and don't know anything<00:03:48.209> about<00:03:48.540> versioning<00:03:49.019> and<00:03:49.200> let's 00:03:49.400 --> 00:03:49.410 align:start position:0% anything about versioning and let's 00:03:49.410 --> 00:03:51.319 align:start position:0% anything about versioning and let's let's<00:03:49.769> instill<00:03:50.160> best<00:03:50.459> practices<00:03:50.970> software 00:03:51.319 --> 00:03:51.329 align:start position:0% let's instill best practices software 00:03:51.329 --> 00:03:52.640 align:start position:0% let's instill best practices software engineering<00:03:51.690> kind<00:03:51.959> of<00:03:52.019> techniques<00:03:52.350> on<00:03:52.500> them 00:03:52.640 --> 00:03:52.650 align:start position:0% engineering kind of techniques on them 00:03:52.650 --> 00:03:54.800 align:start position:0% engineering kind of techniques on them and<00:03:52.980> tell<00:03:53.430> help<00:03:53.970> them<00:03:54.120> understand<00:03:54.540> what<00:03:54.690> it 00:03:54.800 --> 00:03:54.810 align:start position:0% and tell help them understand what it 00:03:54.810 --> 00:03:56.270 align:start position:0% and tell help them understand what it means<00:03:54.959> to<00:03:55.139> put<00:03:55.290> things<00:03:55.410> into<00:03:55.620> production<00:03:55.950> and 00:03:56.270 --> 00:03:56.280 align:start position:0% means to put things into production and 00:03:56.280 --> 00:03:58.039 align:start position:0% means to put things into production and what<00:03:56.370> uptime<00:03:56.700> really<00:03:56.970> is<00:03:57.060> about<00:03:57.329> and<00:03:57.720> so<00:03:57.900> on 00:03:58.039 --> 00:03:58.049 align:start position:0% what uptime really is about and so on 00:03:58.049 --> 00:03:59.900 align:start position:0% what uptime really is about and so on but<00:03:58.680> I<00:03:58.739> think<00:03:58.950> it's<00:03:59.069> also<00:03:59.220> more<00:03:59.489> nuanced<00:03:59.819> than 00:03:59.900 --> 00:03:59.910 align:start position:0% but I think it's also more nuanced than 00:03:59.910 --> 00:04:02.449 align:start position:0% but I think it's also more nuanced than that<00:04:00.060> I<00:04:00.150> do<00:04:00.269> think<00:04:00.720> that<00:04:00.989> you<00:04:01.290> know<00:04:01.410> the<00:04:01.769> just 00:04:02.449 --> 00:04:02.459 align:start position:0% that I do think that you know the just 00:04:02.459 --> 00:04:04.670 align:start position:0% that I do think that you know the just the<00:04:02.639> the<00:04:03.239> process<00:04:03.870> of<00:04:03.989> developing<00:04:04.530> an 00:04:04.670 --> 00:04:04.680 align:start position:0% the the process of developing an 00:04:04.680 --> 00:04:06.650 align:start position:0% the the process of developing an application<00:04:05.100> is<00:04:05.700> quite<00:04:06.060> different<00:04:06.510> than<00:04:06.600> the 00:04:06.650 --> 00:04:06.660 align:start position:0% application is quite different than the 00:04:06.660 --> 00:04:08.090 align:start position:0% application is quite different than the process<00:04:06.870> of<00:04:07.170> developing<00:04:07.590> a<00:04:07.680> traditional 00:04:08.090 --> 00:04:08.100 align:start position:0% process of developing a traditional 00:04:08.100 --> 00:04:09.500 align:start position:0% process of developing a traditional application<00:04:08.639> it's<00:04:09.030> a<00:04:09.090> traditional 00:04:09.500 --> 00:04:09.510 align:start position:0% application it's a traditional 00:04:09.510 --> 00:04:11.330 align:start position:0% application it's a traditional application<00:04:10.049> development<00:04:10.380> you<00:04:10.950> know<00:04:11.069> you<00:04:11.130> hit 00:04:11.330 --> 00:04:11.340 align:start position:0% application development you know you hit 00:04:11.340 --> 00:04:13.729 align:start position:0% application development you know you hit the<00:04:11.459> up-arrow<00:04:11.579> you<00:04:12.209> hit<00:04:12.329> make<00:04:12.540> on<00:04:12.720> GCC<00:04:13.230> and<00:04:13.440> and 00:04:13.729 --> 00:04:13.739 align:start position:0% the up-arrow you hit make on GCC and and 00:04:13.739 --> 00:04:15.410 align:start position:0% the up-arrow you hit make on GCC and and your<00:04:13.920> artifacts<00:04:14.519> build<00:04:14.760> and<00:04:14.940> hopefully<00:04:15.299> your 00:04:15.410 --> 00:04:15.420 align:start position:0% your artifacts build and hopefully your 00:04:15.420 --> 00:04:16.759 align:start position:0% your artifacts build and hopefully your tests<00:04:15.690> are<00:04:15.720> run<00:04:15.870> too<00:04:16.049> and<00:04:16.289> you<00:04:16.530> get<00:04:16.650> all<00:04:16.739> green 00:04:16.759 --> 00:04:16.769 align:start position:0% tests are run too and you get all green 00:04:16.769 --> 00:04:18.170 align:start position:0% tests are run too and you get all green lights<00:04:16.979> and<00:04:17.310> you're<00:04:17.400> like<00:04:17.519> ok<00:04:17.789> I'm<00:04:17.880> ready<00:04:18.030> to 00:04:18.170 --> 00:04:18.180 align:start position:0% lights and you're like ok I'm ready to 00:04:18.180 --> 00:04:20.270 align:start position:0% lights and you're like ok I'm ready to go<00:04:18.329> here<00:04:19.049> you<00:04:19.079> hit<00:04:19.289> the<00:04:19.440> up-arrow<00:04:19.680> you<00:04:19.950> hit<00:04:20.100> go 00:04:20.270 --> 00:04:20.280 align:start position:0% go here you hit the up-arrow you hit go 00:04:20.280 --> 00:04:22.820 align:start position:0% go here you hit the up-arrow you hit go and<00:04:20.639> you<00:04:20.850> wait<00:04:21.030> for<00:04:21.060> 48<00:04:21.840> hours<00:04:21.930> yes<00:04:22.380> your<00:04:22.590> model 00:04:22.820 --> 00:04:22.830 align:start position:0% and you wait for 48 hours yes your model 00:04:22.830 --> 00:04:24.350 align:start position:0% and you wait for 48 hours yes your model to<00:04:22.860> converge<00:04:23.190> all<00:04:23.400> the<00:04:23.490> way<00:04:23.610> yeah<00:04:24.000> and<00:04:24.210> you 00:04:24.350 --> 00:04:24.360 align:start position:0% to converge all the way yeah and you 00:04:24.360 --> 00:04:24.770 align:start position:0% to converge all the way yeah and you don't 00:04:24.770 --> 00:04:24.780 align:start position:0% don't 00:04:24.780 --> 00:04:26.570 align:start position:0% don't if<00:04:24.960> it's<00:04:25.080> actually<00:04:25.410> gonna<00:04:25.500> work<00:04:25.620> it's<00:04:26.250> harder 00:04:26.570 --> 00:04:26.580 align:start position:0% if it's actually gonna work it's harder 00:04:26.580 --> 00:04:28.640 align:start position:0% if it's actually gonna work it's harder to<00:04:26.700> think<00:04:26.910> about<00:04:27.120> what<00:04:27.540> what<00:04:27.990> success<00:04:28.320> is<00:04:28.350> and 00:04:28.640 --> 00:04:28.650 align:start position:0% to think about what what success is and 00:04:28.650 --> 00:04:31.490 align:start position:0% to think about what what success is and what<00:04:29.010> failure<00:04:29.310> is<00:04:29.460> and<00:04:29.670> so<00:04:30.030> I<00:04:30.060> think<00:04:30.180> that<00:04:30.500> in 00:04:31.490 --> 00:04:31.500 align:start position:0% what failure is and so I think that in 00:04:31.500 --> 00:04:33.290 align:start position:0% what failure is and so I think that in in<00:04:31.980> part<00:04:32.220> because<00:04:32.340> of<00:04:32.610> that<00:04:32.700> you've<00:04:32.910> got<00:04:32.940> a<00:04:33.060> lot 00:04:33.290 --> 00:04:33.300 align:start position:0% in part because of that you've got a lot 00:04:33.300 --> 00:04:34.520 align:start position:0% in part because of that you've got a lot more<00:04:33.330> tooling<00:04:33.780> that<00:04:33.810> needs<00:04:34.080> to<00:04:34.200> be<00:04:34.320> developed 00:04:34.520 --> 00:04:34.530 align:start position:0% more tooling that needs to be developed 00:04:34.530 --> 00:04:36.530 align:start position:0% more tooling that needs to be developed to<00:04:34.770> make<00:04:34.920> the<00:04:34.980> process<00:04:35.100> of<00:04:35.550> developing<00:04:36.390> these 00:04:36.530 --> 00:04:36.540 align:start position:0% to make the process of developing these 00:04:36.540 --> 00:04:39.770 align:start position:0% to make the process of developing these applications<00:04:37.110> much<00:04:37.350> more<00:04:37.530> pleasant<00:04:38.000> so<00:04:39.000> a 00:04:39.770 --> 00:04:39.780 align:start position:0% applications much more pleasant so a 00:04:39.780 --> 00:04:42.110 align:start position:0% applications much more pleasant so a little<00:04:40.620> bit<00:04:40.770> you<00:04:40.920> know<00:04:41.040> if<00:04:41.220> you<00:04:41.310> know<00:04:41.460> kind<00:04:42.060> of 00:04:42.110 --> 00:04:42.120 align:start position:0% little bit you know if you know kind of 00:04:42.120 --> 00:04:45.460 align:start position:0% little bit you know if you know kind of what<00:04:42.240> is<00:04:42.390> your<00:04:42.420> like<00:04:43.140> what<00:04:43.890> is<00:04:44.040> your 00:04:45.460 --> 00:04:45.470 align:start position:0% what is your like what is your 00:04:45.470 --> 00:04:47.870 align:start position:0% what is your like what is your dissociating<00:04:46.470> point<00:04:46.770> right<00:04:47.130> I<00:04:47.160> know<00:04:47.430> new 00:04:47.870 --> 00:04:47.880 align:start position:0% dissociating point right I know new 00:04:47.880 --> 00:04:49.430 align:start position:0% dissociating point right I know new companies<00:04:48.270> in<00:04:48.300> the<00:04:48.480> space<00:04:48.570> they<00:04:49.020> come<00:04:49.320> up 00:04:49.430 --> 00:04:49.440 align:start position:0% companies in the space they come up 00:04:49.440 --> 00:04:51.980 align:start position:0% companies in the space they come up regularly<00:04:49.920> so<00:04:50.760> what<00:04:51.120> is<00:04:51.270> your<00:04:51.420> key<00:04:51.570> strength 00:04:51.980 --> 00:04:51.990 align:start position:0% regularly so what is your key strength 00:04:51.990 --> 00:04:53.480 align:start position:0% regularly so what is your key strength and<00:04:52.380> what<00:04:52.530> is<00:04:52.650> your<00:04:52.800> core<00:04:52.950> like<00:04:53.070> you<00:04:53.160> know<00:04:53.220> core 00:04:53.480 --> 00:04:53.490 align:start position:0% and what is your core like you know core 00:04:53.490 --> 00:04:55.400 align:start position:0% and what is your core like you know core competence<00:04:54.030> which<00:04:54.780> makes<00:04:54.990> you<00:04:55.110> think<00:04:55.260> you're 00:04:55.400 --> 00:04:55.410 align:start position:0% competence which makes you think you're 00:04:55.410 --> 00:04:57.170 align:start position:0% competence which makes you think you're gonna<00:04:55.530> succeed<00:04:55.800> and<00:04:56.280> kind<00:04:56.550> of<00:04:56.640> winning<00:04:57.030> this 00:04:57.170 --> 00:04:57.180 align:start position:0% gonna succeed and kind of winning this 00:04:57.180 --> 00:04:59.630 align:start position:0% gonna succeed and kind of winning this in<00:04:57.450> the<00:04:57.570> space<00:04:57.750> yeah<00:04:58.260> so<00:04:58.530> honestly<00:04:59.100> like<00:04:59.280> the 00:04:59.630 --> 00:04:59.640 align:start position:0% in the space yeah so honestly like the 00:04:59.640 --> 00:05:02.990 align:start position:0% in the space yeah so honestly like the team<00:04:59.850> is<00:05:00.090> is<00:05:00.330> a<00:05:00.390> huge<00:05:00.840> piece<00:05:01.380> of<00:05:01.410> this<00:05:01.680> so<00:05:02.000> we've 00:05:02.990 --> 00:05:03.000 align:start position:0% team is is a huge piece of this so we've 00:05:03.000 --> 00:05:05.090 align:start position:0% team is is a huge piece of this so we've got<00:05:03.180> a<00:05:03.210> bunch<00:05:03.390> of<00:05:03.510> people<00:05:03.930> from<00:05:04.170> PhDs<00:05:05.070> from 00:05:05.090 --> 00:05:05.100 align:start position:0% got a bunch of people from PhDs from 00:05:05.100 --> 00:05:09.650 align:start position:0% got a bunch of people from PhDs from places<00:05:05.490> like<00:05:05.520> Berkeley<00:05:06.860> Chicago<00:05:07.860> NYU<00:05:08.660> X 00:05:09.650 --> 00:05:09.660 align:start position:0% places like Berkeley Chicago NYU X 00:05:09.660 --> 00:05:12.380 align:start position:0% places like Berkeley Chicago NYU X schoolers<00:05:10.430> people<00:05:11.430> who've<00:05:11.610> been<00:05:11.760> at<00:05:11.910> places 00:05:12.380 --> 00:05:12.390 align:start position:0% schoolers people who've been at places 00:05:12.390 --> 00:05:13.580 align:start position:0% schoolers people who've been at places like<00:05:12.420> Pinterest<00:05:12.870> and<00:05:13.020> so<00:05:13.140> on<00:05:13.260> so<00:05:13.410> really 00:05:13.580 --> 00:05:13.590 align:start position:0% like Pinterest and so on so really 00:05:13.590 --> 00:05:15.260 align:start position:0% like Pinterest and so on so really strong<00:05:14.040> technical<00:05:14.670> engineering<00:05:15.030> team 00:05:15.260 --> 00:05:15.270 align:start position:0% strong technical engineering team 00:05:15.270 --> 00:05:18.680 align:start position:0% strong technical engineering team experts<00:05:16.140> in<00:05:16.760> distributed<00:05:17.760> systems<00:05:18.150> so<00:05:18.540> my 00:05:18.680 --> 00:05:18.690 align:start position:0% experts in distributed systems so my 00:05:18.690 --> 00:05:20.450 align:start position:0% experts in distributed systems so my co-founder<00:05:18.900> Neal<00:05:19.320> Conway<00:05:19.710> worked<00:05:20.370> on 00:05:20.450 --> 00:05:20.460 align:start position:0% co-founder Neal Conway worked on 00:05:20.460 --> 00:05:22.550 align:start position:0% co-founder Neal Conway worked on mesosphere<00:05:20.790> worked<00:05:21.180> on<00:05:21.330> on<00:05:21.570> Apache<00:05:21.810> Bezos<00:05:22.290> I 00:05:22.550 --> 00:05:22.560 align:start position:0% mesosphere worked on on Apache Bezos I 00:05:22.560 --> 00:05:24.950 align:start position:0% mesosphere worked on on Apache Bezos I myself<00:05:22.950> worked<00:05:23.550> on<00:05:23.669> Apache<00:05:24.300> spark<00:05:24.660> and<00:05:24.900> I 00:05:24.950 --> 00:05:24.960 align:start position:0% myself worked on Apache spark and I 00:05:24.960 --> 00:05:27.920 align:start position:0% myself worked on Apache spark and I lived<00:05:25.169> so<00:05:25.410> on<00:05:25.560> as<00:05:26.520> well<00:05:26.760> as<00:05:26.880> my<00:05:27.000> my<00:05:27.510> co-founder 00:05:27.920 --> 00:05:27.930 align:start position:0% lived so on as well as my my co-founder 00:05:27.930 --> 00:05:30.830 align:start position:0% lived so on as well as my my co-founder me<00:05:28.110> tal<00:05:28.320> Walker<00:05:28.910> Amin<00:05:29.910> is<00:05:30.060> a<00:05:30.090> professor<00:05:30.630> in<00:05:30.750> the 00:05:30.830 --> 00:05:30.840 align:start position:0% me tal Walker Amin is a professor in the 00:05:30.840 --> 00:05:32.270 align:start position:0% me tal Walker Amin is a professor in the machine<00:05:31.080> learning<00:05:31.169> division<00:05:31.500> at<00:05:31.950> Carnegie 00:05:32.270 --> 00:05:32.280 align:start position:0% machine learning division at Carnegie 00:05:32.280 --> 00:05:33.890 align:start position:0% machine learning division at Carnegie Mellon<00:05:32.580> so<00:05:32.730> we've<00:05:32.880> got<00:05:33.000> a<00:05:33.030> really<00:05:33.270> strong<00:05:33.419> core 00:05:33.890 --> 00:05:33.900 align:start position:0% Mellon so we've got a really strong core 00:05:33.900 --> 00:05:35.750 align:start position:0% Mellon so we've got a really strong core of<00:05:34.050> people<00:05:34.169> who<00:05:34.320> both<00:05:34.590> know<00:05:34.830> the<00:05:35.190> distributed 00:05:35.750 --> 00:05:35.760 align:start position:0% of people who both know the distributed 00:05:35.760 --> 00:05:37.340 align:start position:0% of people who both know the distributed systems<00:05:36.090> side<00:05:36.270> of<00:05:36.360> things<00:05:36.570> as<00:05:36.750> well<00:05:36.930> as<00:05:36.960> these 00:05:37.340 --> 00:05:37.350 align:start position:0% systems side of things as well as these 00:05:37.350 --> 00:05:39.110 align:start position:0% systems side of things as well as these scalable<00:05:37.800> machine<00:05:38.070> learning<00:05:38.340> methods<00:05:38.640> so 00:05:39.110 --> 00:05:39.120 align:start position:0% scalable machine learning methods so 00:05:39.120 --> 00:05:40.760 align:start position:0% scalable machine learning methods so that's<00:05:39.270> a<00:05:39.360> big<00:05:39.540> chunk<00:05:39.840> of<00:05:39.960> it<00:05:40.080> from<00:05:40.680> a 00:05:40.760 --> 00:05:40.770 align:start position:0% that's a big chunk of it from a 00:05:40.770 --> 00:05:42.469 align:start position:0% that's a big chunk of it from a technology<00:05:41.070> perspective<00:05:41.430> what<00:05:42.090> we've<00:05:42.240> done 00:05:42.469 --> 00:05:42.479 align:start position:0% technology perspective what we've done 00:05:42.479 --> 00:05:45.409 align:start position:0% technology perspective what we've done is<00:05:42.720> is<00:05:43.020> take<00:05:44.010> a<00:05:44.040> bunch<00:05:44.370> of<00:05:44.520> the<00:05:45.060> research<00:05:45.390> that 00:05:45.409 --> 00:05:45.419 align:start position:0% is is take a bunch of the research that 00:05:45.419 --> 00:05:47.360 align:start position:0% is is take a bunch of the research that we<00:05:45.780> we've<00:05:46.350> been<00:05:46.530> doing<00:05:46.740> over<00:05:46.800> the<00:05:46.919> last<00:05:47.010> call 00:05:47.360 --> 00:05:47.370 align:start position:0% we we've been doing over the last call 00:05:47.370 --> 00:05:50.180 align:start position:0% we we've been doing over the last call at<00:05:47.490> five<00:05:47.640> five<00:05:48.210> plus<00:05:48.240> years<00:05:48.470> around<00:05:49.470> scalable 00:05:50.180 --> 00:05:50.190 align:start position:0% at five five plus years around scalable 00:05:50.190 --> 00:05:53.240 align:start position:0% at five five plus years around scalable methods<00:05:50.280> for<00:05:50.870> ml<00:05:51.870> and<00:05:52.140> and<00:05:52.320> think<00:05:52.710> hard<00:05:53.040> about 00:05:53.240 --> 00:05:53.250 align:start position:0% methods for ml and and think hard about 00:05:53.250 --> 00:05:54.980 align:start position:0% methods for ml and and think hard about how<00:05:53.580> do<00:05:53.910> we<00:05:54.030> commercialize<00:05:54.510> these<00:05:54.660> and<00:05:54.870> how<00:05:54.930> do 00:05:54.980 --> 00:05:54.990 align:start position:0% how do we commercialize these and how do 00:05:54.990 --> 00:05:56.840 align:start position:0% how do we commercialize these and how do we<00:05:55.140> scale<00:05:55.320> them<00:05:55.500> up<00:05:55.590> to<00:05:55.770> the<00:05:55.919> new<00:05:56.070> new<00:05:56.490> world<00:05:56.669> of 00:05:56.840 --> 00:05:56.850 align:start position:0% we scale them up to the new new world of 00:05:56.850 --> 00:05:58.969 align:start position:0% we scale them up to the new new world of deep<00:05:57.000> learning<00:05:57.229> and<00:05:58.229> so<00:05:58.560> we<00:05:58.680> have<00:05:58.830> some 00:05:58.969 --> 00:05:58.979 align:start position:0% deep learning and so we have some 00:05:58.979 --> 00:06:01.340 align:start position:0% deep learning and so we have some proprietary<00:05:59.190> algorithms<00:06:00.000> around<00:06:00.360> a<00:06:00.960> type<00:06:01.320> of 00:06:01.340 --> 00:06:01.350 align:start position:0% proprietary algorithms around a type of 00:06:01.350 --> 00:06:03.590 align:start position:0% proprietary algorithms around a type of revenue<00:06:01.620> optimization<00:06:02.400> I'll<00:06:02.580> be<00:06:03.270> telling<00:06:03.450> you 00:06:03.590 --> 00:06:03.600 align:start position:0% revenue optimization I'll be telling you 00:06:03.600 --> 00:06:04.880 align:start position:0% revenue optimization I'll be telling you about<00:06:03.660> sort<00:06:04.020> of<00:06:04.080> the<00:06:04.169> open<00:06:04.410> components<00:06:04.830> of 00:06:04.880 --> 00:06:04.890 align:start position:0% about sort of the open components of 00:06:04.890 --> 00:06:07.850 align:start position:0% about sort of the open components of that<00:06:04.919> later<00:06:05.220> tonight<00:06:05.540> we<00:06:06.540> have<00:06:06.690> some<00:06:07.560> really 00:06:07.850 --> 00:06:07.860 align:start position:0% that later tonight we have some really 00:06:07.860 --> 00:06:09.710 align:start position:0% that later tonight we have some really sophisticated<00:06:08.430> techniques<00:06:09.240> around<00:06:09.360> cost 00:06:09.710 --> 00:06:09.720 align:start position:0% sophisticated techniques around cost 00:06:09.720 --> 00:06:12.650 align:start position:0% sophisticated techniques around cost modeling<00:06:10.140> for<00:06:10.700> distributed<00:06:11.700> training<00:06:12.450> and 00:06:12.650 --> 00:06:12.660 align:start position:0% modeling for distributed training and 00:06:12.660 --> 00:06:15.890 align:start position:0% modeling for distributed training and deployment<00:06:13.500> and<00:06:13.919> so<00:06:14.580> all<00:06:14.970> of<00:06:15.150> that<00:06:15.180> kind<00:06:15.870> of 00:06:15.890 --> 00:06:15.900 align:start position:0% deployment and so all of that kind of 00:06:15.900 --> 00:06:18.500 align:start position:0% deployment and so all of that kind of adds<00:06:16.110> up<00:06:16.350> to<00:06:16.430> an<00:06:17.430> end-to-end<00:06:17.790> experience<00:06:18.030> we 00:06:18.500 --> 00:06:18.510 align:start position:0% adds up to an end-to-end experience we 00:06:18.510 --> 00:06:20.210 align:start position:0% adds up to an end-to-end experience we think<00:06:18.540> in<00:06:18.960> that<00:06:19.110> hopefully<00:06:19.740> our<00:06:19.830> customers 00:06:20.210 --> 00:06:20.220 align:start position:0% think in that hopefully our customers 00:06:20.220 --> 00:06:23.240 align:start position:0% think in that hopefully our customers think<00:06:20.990> results<00:06:21.990> in<00:06:22.110> in<00:06:22.500> dramatically<00:06:23.070> faster 00:06:23.240 --> 00:06:23.250 align:start position:0% think results in in dramatically faster 00:06:23.250 --> 00:06:25.690 align:start position:0% think results in in dramatically faster times<00:06:23.760> to<00:06:23.970> build<00:06:24.210> and<00:06:24.660> ship<00:06:24.840> applications 00:06:25.690 --> 00:06:25.700 align:start position:0% times to build and ship applications 00:06:25.700 --> 00:06:28.159 align:start position:0% times to build and ship applications unlike<00:06:26.700> you<00:06:27.150> know<00:06:27.270> you've<00:06:27.479> ever<00:06:27.630> seen<00:06:27.840> before 00:06:28.159 --> 00:06:28.169 align:start position:0% unlike you know you've ever seen before 00:06:28.169 --> 00:06:29.999 align:start position:0% unlike you know you've ever seen before nice<00:06:28.800> nice<00:06:29.220> I<00:06:29.460> mean 00:06:29.999 --> 00:06:30.009 align:start position:0% nice nice I mean 00:06:30.009 --> 00:06:31.290 align:start position:0% nice nice I mean real<00:06:30.189> excited<00:06:30.520> that<00:06:30.639> you<00:06:30.729> know<00:06:30.789> you<00:06:30.939> met<00:06:31.119> them 00:06:31.290 --> 00:06:31.300 align:start position:0% real excited that you know you met them 00:06:31.300 --> 00:06:33.209 align:start position:0% real excited that you know you met them that<00:06:31.479> the<00:06:31.599> citizen<00:06:32.110> component<00:06:32.650> because<00:06:32.860> often 00:06:33.209 --> 00:06:33.219 align:start position:0% that the citizen component because often 00:06:33.219 --> 00:06:36.659 align:start position:0% that the citizen component because often in<00:06:33.789> kind<00:06:34.150> of<00:06:34.210> traditional<00:06:34.839> AI<00:06:35.199> talks<00:06:36.129> we<00:06:36.430> we 00:06:36.659 --> 00:06:36.669 align:start position:0% in kind of traditional AI talks we we 00:06:36.669 --> 00:06:38.579 align:start position:0% in kind of traditional AI talks we we see<00:06:36.789> a<00:06:36.819> graph<00:06:37.059> we<00:06:37.300> seen<00:06:37.539> on<00:06:37.779> a<00:06:37.900> notebook<00:06:38.110> I 00:06:38.579 --> 00:06:38.589 align:start position:0% see a graph we seen on a notebook I 00:06:38.589 --> 00:06:40.049 align:start position:0% see a graph we seen on a notebook I don't<00:06:39.159> see<00:06:39.249> the<00:06:39.369> disturbance<00:06:39.759> distance 00:06:40.049 --> 00:06:40.059 align:start position:0% don't see the disturbance distance 00:06:40.059 --> 00:06:42.029 align:start position:0% don't see the disturbance distance behind<00:06:40.210> this<00:06:40.539> and<00:06:40.839> I'm<00:06:41.529> really<00:06:41.740> excited<00:06:41.919> that 00:06:42.029 --> 00:06:42.039 align:start position:0% behind this and I'm really excited that 00:06:42.039 --> 00:06:43.499 align:start position:0% behind this and I'm really excited that you<00:06:42.219> know<00:06:42.249> that<00:06:42.639> that<00:06:42.759> comes<00:06:42.969> with<00:06:43.089> the<00:06:43.149> forum 00:06:43.499 --> 00:06:43.509 align:start position:0% you know that that comes with the forum 00:06:43.509 --> 00:06:45.420 align:start position:0% you know that that comes with the forum because<00:06:43.599> in<00:06:43.839> a<00:06:43.930> real<00:06:44.110> production<00:06:44.379> setting<00:06:44.889> you 00:06:45.420 --> 00:06:45.430 align:start position:0% because in a real production setting you 00:06:45.430 --> 00:06:47.760 align:start position:0% because in a real production setting you need<00:06:45.639> any<00:06:46.330> of<00:06:46.389> the<00:06:46.479> six<00:06:46.659> release<00:06:46.990> yeah<00:06:47.529> it's 00:06:47.760 --> 00:06:47.770 align:start position:0% need any of the six release yeah it's 00:06:47.770 --> 00:06:49.920 align:start position:0% need any of the six release yeah it's it's<00:06:48.099> really<00:06:48.339> hard<00:06:48.639> I<00:06:48.789> mean<00:06:48.909> it<00:06:49.089> you<00:06:49.659> know<00:06:49.689> if 00:06:49.920 --> 00:06:49.930 align:start position:0% it's really hard I mean it you know if 00:06:49.930 --> 00:06:51.329 align:start position:0% it's really hard I mean it you know if you're<00:06:49.960> talking<00:06:50.229> about<00:06:50.499> waiting<00:06:50.830> you<00:06:51.249> know 00:06:51.329 --> 00:06:51.339 align:start position:0% you're talking about waiting you know 00:06:51.339 --> 00:06:54.119 align:start position:0% you're talking about waiting you know days<00:06:51.639> or<00:06:51.789> weeks<00:06:51.939> for<00:06:52.149> models<00:06:52.419> to<00:06:52.449> converge<00:06:53.129> you 00:06:54.119 --> 00:06:54.129 align:start position:0% days or weeks for models to converge you 00:06:54.129 --> 00:06:58.920 align:start position:0% days or weeks for models to converge you are<00:06:54.520> starting<00:06:55.300> to<00:06:55.360> think<00:06:55.479> about<00:06:57.689> one<00:06:58.689> of<00:06:58.719> the 00:06:58.920 --> 00:06:58.930 align:start position:0% are starting to think about one of the 00:06:58.930 --> 00:07:00.359 align:start position:0% are starting to think about one of the natural<00:06:59.080> things<00:06:59.439> that<00:06:59.499> you<00:06:59.680> think<00:06:59.889> about<00:07:00.069> is 00:07:00.359 --> 00:07:00.369 align:start position:0% natural things that you think about is 00:07:00.369 --> 00:07:02.519 align:start position:0% natural things that you think about is how<00:07:00.909> do<00:07:00.969> I<00:07:01.059> throw<00:07:01.330> more<00:07:01.539> GPUs<00:07:01.960> at<00:07:02.110> this<00:07:02.259> than<00:07:02.439> I 00:07:02.519 --> 00:07:02.529 align:start position:0% how do I throw more GPUs at this than I 00:07:02.529 --> 00:07:04.829 align:start position:0% how do I throw more GPUs at this than I then<00:07:02.949> just<00:07:03.279> the<00:07:03.370> one<00:07:03.550> at<00:07:03.699> one<00:07:03.969> at<00:07:04.089> a<00:07:04.149> time<00:07:04.360> how 00:07:04.829 --> 00:07:04.839 align:start position:0% then just the one at one at a time how 00:07:04.839 --> 00:07:05.939 align:start position:0% then just the one at one at a time how do<00:07:04.899> I<00:07:05.020> do<00:07:05.080> that<00:07:05.289> in<00:07:05.439> an<00:07:05.559> efficient<00:07:05.919> way 00:07:05.939 --> 00:07:05.949 align:start position:0% do I do that in an efficient way 00:07:05.949 --> 00:07:07.679 align:start position:0% do I do that in an efficient way communication<00:07:06.789> efficient<00:07:07.180> way<00:07:07.270> how<00:07:07.449> do<00:07:07.509> I<00:07:07.629> get 00:07:07.679 --> 00:07:07.689 align:start position:0% communication efficient way how do I get 00:07:07.689 --> 00:07:09.600 align:start position:0% communication efficient way how do I get scalability<00:07:08.469> that<00:07:08.949> I<00:07:09.129> wouldn't<00:07:09.430> get 00:07:09.600 --> 00:07:09.610 align:start position:0% scalability that I wouldn't get 00:07:09.610 --> 00:07:10.070 align:start position:0% scalability that I wouldn't get otherwise 00:07:10.070 --> 00:07:10.080 align:start position:0% otherwise 00:07:10.080 --> 00:07:12.989 align:start position:0% otherwise from<00:07:11.080> from<00:07:11.469> sort<00:07:11.680> of<00:07:11.710> the<00:07:11.919> the<00:07:12.249> office<00:07:12.550> shelf 00:07:12.989 --> 00:07:12.999 align:start position:0% from from sort of the the office shelf 00:07:12.999 --> 00:07:14.369 align:start position:0% from from sort of the the office shelf tools<00:07:13.210> how<00:07:13.629> do<00:07:13.719> I<00:07:13.839> do<00:07:13.899> that<00:07:14.080> in<00:07:14.199> a<00:07:14.259> very 00:07:14.369 --> 00:07:14.379 align:start position:0% tools how do I do that in a very 00:07:14.379 --> 00:07:15.540 align:start position:0% tools how do I do that in a very principled<00:07:14.860> way<00:07:14.979> how<00:07:15.189> do<00:07:15.249> I<00:07:15.339> make<00:07:15.520> it 00:07:15.540 --> 00:07:15.550 align:start position:0% principled way how do I make it 00:07:15.550 --> 00:07:17.219 align:start position:0% principled way how do I make it fault-tolerant<00:07:16.330> how<00:07:16.629> do<00:07:16.689> I<00:07:16.779> make<00:07:16.960> sure<00:07:16.990> it's 00:07:17.219 --> 00:07:17.229 align:start position:0% fault-tolerant how do I make sure it's 00:07:17.229 --> 00:07:19.649 align:start position:0% fault-tolerant how do I make sure it's versioned<00:07:17.649> and<00:07:17.979> so<00:07:18.159> on<00:07:18.309> and<00:07:18.550> and<00:07:18.939> that's<00:07:19.419> where 00:07:19.649 --> 00:07:19.659 align:start position:0% versioned and so on and and that's where 00:07:19.659 --> 00:07:21.209 align:start position:0% versioned and so on and and that's where sort<00:07:19.930> of<00:07:19.990> the<00:07:20.050> system's<00:07:20.379> expertise<00:07:20.830> starts<00:07:21.099> to 00:07:21.209 --> 00:07:21.219 align:start position:0% sort of the system's expertise starts to 00:07:21.219 --> 00:07:22.889 align:start position:0% sort of the system's expertise starts to come<00:07:21.399> in<00:07:21.580> and<00:07:21.729> I<00:07:22.120> think<00:07:22.330> that's<00:07:22.509> a<00:07:22.659> really 00:07:22.889 --> 00:07:22.899 align:start position:0% come in and I think that's a really 00:07:22.899 --> 00:07:25.499 align:start position:0% come in and I think that's a really important<00:07:23.610> component<00:07:24.610> of<00:07:24.939> the<00:07:25.029> experience 00:07:25.499 --> 00:07:25.509 align:start position:0% important component of the experience 00:07:25.509 --> 00:07:28.019 align:start position:0% important component of the experience yes<00:07:25.930> so<00:07:26.439> this<00:07:26.589> is<00:07:26.680> great<00:07:26.860> so<00:07:27.129> I<00:07:27.339> know<00:07:27.550> that<00:07:27.699> you 00:07:28.019 --> 00:07:28.029 align:start position:0% yes so this is great so I know that you 00:07:28.029 --> 00:07:31.260 align:start position:0% yes so this is great so I know that you know<00:07:28.149> I'm<00:07:28.719> blab<00:07:29.050> actually<00:07:29.469> makes<00:07:30.059> the<00:07:31.059> same 00:07:31.260 --> 00:07:31.270 align:start position:0% know I'm blab actually makes the same 00:07:31.270 --> 00:07:34.529 align:start position:0% know I'm blab actually makes the same kind<00:07:31.659> of<00:07:31.830> connection<00:07:32.830> right<00:07:33.699> but<00:07:34.029> maybe<00:07:34.180> for 00:07:34.529 --> 00:07:34.539 align:start position:0% kind of connection right but maybe for 00:07:34.539 --> 00:07:38.369 align:start position:0% kind of connection right but maybe for the<00:07:34.629> reviewers<00:07:34.959> who<00:07:35.259> did<00:07:35.680> not<00:07:37.199> read<00:07:38.199> up<00:07:38.289> on 00:07:38.369 --> 00:07:38.379 align:start position:0% the reviewers who did not read up on 00:07:38.379 --> 00:07:41.489 align:start position:0% the reviewers who did not read up on that<00:07:38.559> much<00:07:39.039> you<00:07:40.029> can<00:07:40.209> tell<00:07:40.419> us<00:07:40.569> you<00:07:41.139> know<00:07:41.229> what 00:07:41.489 --> 00:07:41.499 align:start position:0% that much you can tell us you know what 00:07:41.499 --> 00:07:44.309 align:start position:0% that much you can tell us you know what makes<00:07:41.709> us<00:07:42.039> special<00:07:42.490> why<00:07:43.059> you<00:07:43.120> know<00:07:43.539> things 00:07:44.309 --> 00:07:44.319 align:start position:0% makes us special why you know things 00:07:44.319 --> 00:07:45.480 align:start position:0% makes us special why you know things like<00:07:44.379> spark<00:07:44.770> came<00:07:44.889> out<00:07:45.039> of<00:07:45.189> it 00:07:45.480 --> 00:07:45.490 align:start position:0% like spark came out of it 00:07:45.490 --> 00:07:47.999 align:start position:0% like spark came out of it yeah<00:07:45.939> so<00:07:46.149> the<00:07:46.269> Empire<00:07:46.539> business<00:07:47.009> terrific 00:07:47.999 --> 00:07:48.009 align:start position:0% yeah so the Empire business terrific 00:07:48.009 --> 00:07:50.579 align:start position:0% yeah so the Empire business terrific place<00:07:48.219> at<00:07:48.399> at<00:07:48.669> a<00:07:48.699> point<00:07:48.909> in<00:07:49.089> time<00:07:49.300> where<00:07:49.659> you 00:07:50.579 --> 00:07:50.589 align:start position:0% place at at a point in time where you 00:07:50.589 --> 00:07:53.399 align:start position:0% place at at a point in time where you know<00:07:50.979> you<00:07:51.039> started<00:07:51.819> to<00:07:51.969> see<00:07:52.180> an<00:07:52.539> industrially 00:07:53.399 --> 00:07:53.409 align:start position:0% know you started to see an industrially 00:07:53.409 --> 00:07:55.320 align:start position:0% know you started to see an industrially this<00:07:53.589> trend<00:07:53.889> where<00:07:54.099> datasets<00:07:54.909> were<00:07:55.059> getting 00:07:55.320 --> 00:07:55.330 align:start position:0% this trend where datasets were getting 00:07:55.330 --> 00:07:57.089 align:start position:0% this trend where datasets were getting bigger<00:07:55.509> and<00:07:55.689> bigger<00:07:55.749> it<00:07:56.110> had<00:07:56.229> never<00:07:56.349> gotten<00:07:56.860> it 00:07:57.089 --> 00:07:57.099 align:start position:0% bigger and bigger it had never gotten it 00:07:57.099 --> 00:07:59.489 align:start position:0% bigger and bigger it had never gotten it had<00:07:57.430> never<00:07:57.550> been<00:07:57.669> nearly<00:07:58.360> as<00:07:58.539> cheap<00:07:58.839> or<00:07:59.050> easy 00:07:59.489 --> 00:07:59.499 align:start position:0% had never been nearly as cheap or easy 00:07:59.499 --> 00:08:01.769 align:start position:0% had never been nearly as cheap or easy to<00:07:59.830> store<00:08:00.189> and<00:08:00.249> and<00:08:00.819> process<00:08:01.269> a<00:08:01.300> lot<00:08:01.509> of<00:08:01.599> data 00:08:01.769 --> 00:08:01.779 align:start position:0% to store and and process a lot of data 00:08:01.779 --> 00:08:04.019 align:start position:0% to store and and process a lot of data but<00:08:02.050> then<00:08:02.800> the<00:08:02.919> question<00:08:03.069> was<00:08:03.430> like<00:08:03.639> alright 00:08:04.019 --> 00:08:04.029 align:start position:0% but then the question was like alright 00:08:04.029 --> 00:08:05.670 align:start position:0% but then the question was like alright we've<00:08:04.209> got<00:08:04.360> all<00:08:04.449> this<00:08:04.569> stuff<00:08:04.749> sitting<00:08:04.959> in<00:08:05.110> HDFS 00:08:05.670 --> 00:08:05.680 align:start position:0% we've got all this stuff sitting in HDFS 00:08:05.680 --> 00:08:08.579 align:start position:0% we've got all this stuff sitting in HDFS what<00:08:06.339> do<00:08:06.399> we<00:08:06.490> do<00:08:06.610> with<00:08:06.729> it<00:08:06.789> and<00:08:07.379> luckily<00:08:08.379> it 00:08:08.579 --> 00:08:08.589 align:start position:0% what do we do with it and luckily it 00:08:08.589 --> 00:08:10.889 align:start position:0% what do we do with it and luckily it turns<00:08:09.099> out<00:08:09.309> that<00:08:09.490> more<00:08:09.939> data<00:08:10.149> is<00:08:10.360> great<00:08:10.629> fuel 00:08:10.889 --> 00:08:10.899 align:start position:0% turns out that more data is great fuel 00:08:10.899 --> 00:08:13.110 align:start position:0% turns out that more data is great fuel for<00:08:11.199> machine<00:08:11.740> learning<00:08:11.769> techniques<00:08:12.399> and<00:08:12.610> so 00:08:13.110 --> 00:08:13.120 align:start position:0% for machine learning techniques and so 00:08:13.120 --> 00:08:15.239 align:start position:0% for machine learning techniques and so the<00:08:13.629> the<00:08:13.959> people<00:08:14.199> who<00:08:14.319> who<00:08:14.709> founded<00:08:15.009> the<00:08:15.129> amp 00:08:15.239 --> 00:08:15.249 align:start position:0% the the people who who founded the amp 00:08:15.249 --> 00:08:17.309 align:start position:0% the the people who who founded the amp lab<00:08:15.459> my<00:08:15.669> adviser<00:08:15.909> Mike<00:08:16.180> Franklin<00:08:16.629> Mike<00:08:16.959> Jordan 00:08:17.309 --> 00:08:17.319 align:start position:0% lab my adviser Mike Franklin Mike Jordan 00:08:17.319 --> 00:08:19.679 align:start position:0% lab my adviser Mike Franklin Mike Jordan who<00:08:18.069> we<00:08:18.189> worked<00:08:18.430> with<00:08:18.580> in<00:08:18.699> the<00:08:18.759> past<00:08:18.819> and<00:08:19.330> and<00:08:19.599> a 00:08:19.679 --> 00:08:19.689 align:start position:0% who we worked with in the past and and a 00:08:19.689 --> 00:08:21.179 align:start position:0% who we worked with in the past and and a bunch<00:08:19.870> of<00:08:19.930> the<00:08:20.050> other<00:08:20.080> you<00:08:20.589> know<00:08:20.620> faculty 00:08:21.179 --> 00:08:21.189 align:start position:0% bunch of the other you know faculty 00:08:21.189 --> 00:08:23.369 align:start position:0% bunch of the other you know faculty there<00:08:21.399> really<00:08:21.849> saw<00:08:22.120> this<00:08:22.269> issue<00:08:22.449> and<00:08:22.719> said<00:08:22.930> hey 00:08:23.369 --> 00:08:23.379 align:start position:0% there really saw this issue and said hey 00:08:23.379 --> 00:08:25.290 align:start position:0% there really saw this issue and said hey wouldn't<00:08:23.740> this<00:08:23.860> be<00:08:23.949> a<00:08:24.099> great<00:08:24.279> opportunity<00:08:24.639> for 00:08:25.290 --> 00:08:25.300 align:start position:0% wouldn't this be a great opportunity for 00:08:25.300 --> 00:08:26.610 align:start position:0% wouldn't this be a great opportunity for communities<00:08:26.050> that<00:08:26.199> haven't<00:08:26.409> traditionally 00:08:26.610 --> 00:08:26.620 align:start position:0% communities that haven't traditionally 00:08:26.620 --> 00:08:28.110 align:start position:0% communities that haven't traditionally talked<00:08:27.159> to<00:08:27.249> each<00:08:27.370> other<00:08:27.519> you<00:08:27.999> know 00:08:28.110 --> 00:08:28.120 align:start position:0% talked to each other you know 00:08:28.120 --> 00:08:30.420 align:start position:0% talked to each other you know distributed<00:08:28.749> systems<00:08:29.110> databases<00:08:29.709> operating 00:08:30.420 --> 00:08:30.430 align:start position:0% distributed systems databases operating 00:08:30.430 --> 00:08:33.449 align:start position:0% distributed systems databases operating systems<00:08:30.789> folks<00:08:31.089> and<00:08:31.680> really<00:08:32.680> hard<00:08:33.039> core<00:08:33.219> deep 00:08:33.449 --> 00:08:33.459 align:start position:0% systems folks and really hard core deep 00:08:33.459 --> 00:08:36.000 align:start position:0% systems folks and really hard core deep theoretical<00:08:33.969> ml<00:08:34.479> folks<00:08:34.719> put<00:08:35.529> them<00:08:35.680> in<00:08:35.769> a<00:08:35.829> room 00:08:36.000 --> 00:08:36.010 align:start position:0% theoretical ml folks put them in a room 00:08:36.010 --> 00:08:37.439 align:start position:0% theoretical ml folks put them in a room for<00:08:36.219> five<00:08:36.430> years<00:08:36.459> and<00:08:36.789> get<00:08:36.940> them<00:08:37.029> talking<00:08:37.180> to 00:08:37.439 --> 00:08:37.449 align:start position:0% for five years and get them talking to 00:08:37.449 --> 00:08:39.480 align:start position:0% for five years and get them talking to each<00:08:37.599> other<00:08:37.719> and<00:08:37.930> this<00:08:38.050> is<00:08:38.199> you<00:08:38.349> know<00:08:38.490> probably 00:08:39.480 --> 00:08:39.490 align:start position:0% each other and this is you know probably 00:08:39.490 --> 00:08:41.760 align:start position:0% each other and this is you know probably 70<00:08:40.180> or<00:08:40.269> 80<00:08:40.389> PhD<00:08:40.870> students<00:08:41.349> you 00:08:41.760 --> 00:08:41.770 align:start position:0% 70 or 80 PhD students you 00:08:41.770 --> 00:08:45.060 align:start position:0% 70 or 80 PhD students you doesn't<00:08:42.070> a<00:08:42.159> dozen<00:08:42.820> or<00:08:42.909> more<00:08:42.940> postdocs<00:08:43.479> a<00:08:44.070> dozen 00:08:45.060 --> 00:08:45.070 align:start position:0% doesn't a dozen or more postdocs a dozen 00:08:45.070 --> 00:08:46.800 align:start position:0% doesn't a dozen or more postdocs a dozen or<00:08:45.160> so<00:08:45.190> faculty<00:08:45.730> members<00:08:46.029> sat<00:08:46.600> there<00:08:46.779> and 00:08:46.800 --> 00:08:46.810 align:start position:0% or so faculty members sat there and 00:08:46.810 --> 00:08:48.269 align:start position:0% or so faculty members sat there and really<00:08:47.380> started<00:08:47.649> to<00:08:47.709> work<00:08:47.830> through<00:08:48.190> these 00:08:48.269 --> 00:08:48.279 align:start position:0% really started to work through these 00:08:48.279 --> 00:08:50.280 align:start position:0% really started to work through these problems<00:08:48.339> and<00:08:48.940> you<00:08:49.540> know<00:08:49.660> I<00:08:49.690> would<00:08:49.899> say<00:08:50.080> that 00:08:50.280 --> 00:08:50.290 align:start position:0% problems and you know I would say that 00:08:50.290 --> 00:08:52.350 align:start position:0% problems and you know I would say that it<00:08:50.440> at<00:08:50.740> the<00:08:50.890> beginning<00:08:51.190> of<00:08:51.339> this<00:08:51.520> this<00:08:51.970> period 00:08:52.350 --> 00:08:52.360 align:start position:0% it at the beginning of this this period 00:08:52.360 --> 00:08:53.730 align:start position:0% it at the beginning of this this period which<00:08:52.510> is<00:08:52.630> when<00:08:52.779> I<00:08:52.810> kind<00:08:53.410> of<00:08:53.470> joined<00:08:53.649> the 00:08:53.730 --> 00:08:53.740 align:start position:0% which is when I kind of joined the 00:08:53.740 --> 00:08:55.949 align:start position:0% which is when I kind of joined the ampulla<00:08:54.040> things<00:08:54.760> were<00:08:54.940> really<00:08:55.630> just<00:08:55.750> getting 00:08:55.949 --> 00:08:55.959 align:start position:0% ampulla things were really just getting 00:08:55.959 --> 00:08:57.389 align:start position:0% ampulla things were really just getting off<00:08:56.140> the<00:08:56.290> ground<00:08:56.320> and<00:08:56.770> people<00:08:56.890> were<00:08:57.100> kind<00:08:57.250> of 00:08:57.389 --> 00:08:57.399 align:start position:0% off the ground and people were kind of 00:08:57.399 --> 00:08:58.860 align:start position:0% off the ground and people were kind of uncomfortable<00:08:57.700> talking<00:08:58.029> to<00:08:58.330> each<00:08:58.480> other 00:08:58.860 --> 00:08:58.870 align:start position:0% uncomfortable talking to each other 00:08:58.870 --> 00:09:00.750 align:start position:0% uncomfortable talking to each other John<00:08:59.230> du<00:08:59.350> cheese<00:08:59.529> sitting<00:08:59.800> in<00:08:59.860> one<00:09:00.010> seat<00:09:00.279> Matea 00:09:00.750 --> 00:09:00.760 align:start position:0% John du cheese sitting in one seat Matea 00:09:00.760 --> 00:09:02.790 align:start position:0% John du cheese sitting in one seat Matea Hari<00:09:01.029> is<00:09:01.149> sitting<00:09:01.360> in<00:09:01.450> another<00:09:01.540> seat<00:09:01.930> and<00:09:02.260> both 00:09:02.790 --> 00:09:02.800 align:start position:0% Hari is sitting in another seat and both 00:09:02.800 --> 00:09:04.320 align:start position:0% Hari is sitting in another seat and both professors<00:09:03.279> at<00:09:03.430> Stanford<00:09:03.820> now<00:09:03.970> but<00:09:04.149> it<00:09:04.240> was 00:09:04.320 --> 00:09:04.330 align:start position:0% professors at Stanford now but it was 00:09:04.330 --> 00:09:05.370 align:start position:0% professors at Stanford now but it was like<00:09:04.510> they<00:09:04.630> never<00:09:04.870> would<00:09:05.020> have<00:09:05.080> talked<00:09:05.290> before 00:09:05.370 --> 00:09:05.380 align:start position:0% like they never would have talked before 00:09:05.380 --> 00:09:08.550 align:start position:0% like they never would have talked before and<00:09:06.180> suddenly<00:09:07.180> they<00:09:07.600> got<00:09:07.750> together<00:09:07.930> and<00:09:08.320> and 00:09:08.550 --> 00:09:08.560 align:start position:0% and suddenly they got together and and 00:09:08.560 --> 00:09:10.380 align:start position:0% and suddenly they got together and and more<00:09:09.070> folks<00:09:09.310> like<00:09:09.550> like<00:09:09.880> them<00:09:10.089> got<00:09:10.240> together 00:09:10.380 --> 00:09:10.390 align:start position:0% more folks like like them got together 00:09:10.390 --> 00:09:12.090 align:start position:0% more folks like like them got together and<00:09:10.660> started<00:09:10.779> thinking<00:09:11.110> about<00:09:11.350> how<00:09:11.890> did<00:09:12.040> the 00:09:12.090 --> 00:09:12.100 align:start position:0% and started thinking about how did the 00:09:12.100 --> 00:09:13.949 align:start position:0% and started thinking about how did the how<00:09:12.399> do<00:09:12.459> we<00:09:12.580> reconcile<00:09:13.029> these<00:09:13.209> two<00:09:13.270> worlds<00:09:13.810> and 00:09:13.949 --> 00:09:13.959 align:start position:0% how do we reconcile these two worlds and 00:09:13.959 --> 00:09:15.720 align:start position:0% how do we reconcile these two worlds and and<00:09:14.200> I<00:09:14.290> think<00:09:14.500> the<00:09:14.589> net<00:09:14.770> effect<00:09:15.160> was<00:09:15.339> something 00:09:15.720 --> 00:09:15.730 align:start position:0% and I think the net effect was something 00:09:15.730 --> 00:09:17.670 align:start position:0% and I think the net effect was something much<00:09:16.180> greater<00:09:16.420> than<00:09:16.750> then<00:09:17.080> either<00:09:17.260> of<00:09:17.470> the<00:09:17.560> two 00:09:17.670 --> 00:09:17.680 align:start position:0% much greater than then either of the two 00:09:17.680 --> 00:09:18.690 align:start position:0% much greater than then either of the two sides<00:09:17.890> would<00:09:18.040> have<00:09:18.130> come<00:09:18.250> up<00:09:18.370> with<00:09:18.490> on<00:09:18.610> their 00:09:18.690 --> 00:09:18.700 align:start position:0% sides would have come up with on their 00:09:18.700 --> 00:09:21.389 align:start position:0% sides would have come up with on their own<00:09:18.880> and<00:09:19.360> it<00:09:19.540> stood<00:09:19.750> for<00:09:19.959> those<00:09:20.709> machines<00:09:21.130> and 00:09:21.389 --> 00:09:21.399 align:start position:0% own and it stood for those machines and 00:09:21.399 --> 00:09:24.180 align:start position:0% own and it stood for those machines and people<00:09:21.490> yeah<00:09:21.940> that's<00:09:22.149> right<00:09:22.980> people<00:09:23.980> was 00:09:24.180 --> 00:09:24.190 align:start position:0% people yeah that's right people was 00:09:24.190 --> 00:09:25.800 align:start position:0% people yeah that's right people was absolutely<00:09:24.610> absolutely<00:09:24.970> a<00:09:25.089> part<00:09:25.270> of<00:09:25.420> this<00:09:25.540> and 00:09:25.800 --> 00:09:25.810 align:start position:0% absolutely absolutely a part of this and 00:09:25.810 --> 00:09:27.750 align:start position:0% absolutely absolutely a part of this and and<00:09:26.050> both<00:09:26.500> sort<00:09:26.770> of<00:09:26.830> the<00:09:26.890> people<00:09:27.100> who<00:09:27.370> are 00:09:27.750 --> 00:09:27.760 align:start position:0% and both sort of the people who are 00:09:27.760 --> 00:09:29.310 align:start position:0% and both sort of the people who are thinking<00:09:28.060> about<00:09:28.360> these<00:09:28.510> things<00:09:28.570> and<00:09:28.899> using 00:09:29.310 --> 00:09:29.320 align:start position:0% thinking about these things and using 00:09:29.320 --> 00:09:31.320 align:start position:0% thinking about these things and using the<00:09:29.500> systems<00:09:29.860> and<00:09:30.100> that<00:09:30.339> was<00:09:30.490> important<00:09:30.910> but 00:09:31.320 --> 00:09:31.330 align:start position:0% the systems and that was important but 00:09:31.330 --> 00:09:33.630 align:start position:0% the systems and that was important but also<00:09:31.510> there<00:09:31.930> was<00:09:32.020> this<00:09:32.170> this<00:09:32.950> really<00:09:33.370> great 00:09:33.630 --> 00:09:33.640 align:start position:0% also there was this this really great 00:09:33.640 --> 00:09:35.579 align:start position:0% also there was this this really great flavor<00:09:34.000> of<00:09:34.270> crowdsourcing<00:09:35.170> and<00:09:35.290> how<00:09:35.410> do<00:09:35.470> we 00:09:35.579 --> 00:09:35.589 align:start position:0% flavor of crowdsourcing and how do we 00:09:35.589 --> 00:09:38.639 align:start position:0% flavor of crowdsourcing and how do we think<00:09:35.920> about<00:09:36.010> how<00:09:36.610> we<00:09:36.670> leverage<00:09:37.020> you<00:09:38.020> know<00:09:38.110> the 00:09:38.639 --> 00:09:38.649 align:start position:0% think about how we leverage you know the 00:09:38.649 --> 00:09:42.030 align:start position:0% think about how we leverage you know the ability<00:09:39.190> to<00:09:39.520> collect<00:09:40.470> human<00:09:41.470> powered<00:09:41.800> data 00:09:42.030 --> 00:09:42.040 align:start position:0% ability to collect human powered data 00:09:42.040 --> 00:09:44.910 align:start position:0% ability to collect human powered data from<00:09:42.339> lots<00:09:42.520> of<00:09:42.670> people<00:09:42.970> at<00:09:43.149> once<00:09:43.620> as<00:09:44.620> part<00:09:44.860> of 00:09:44.910 --> 00:09:44.920 align:start position:0% from lots of people at once as part of 00:09:44.920 --> 00:09:45.840 align:start position:0% from lots of people at once as part of this<00:09:45.040> and<00:09:45.250> so<00:09:45.339> there<00:09:45.490> were<00:09:45.580> some<00:09:45.790> really 00:09:45.840 --> 00:09:45.850 align:start position:0% this and so there were some really 00:09:45.850 --> 00:09:47.850 align:start position:0% this and so there were some really interesting<00:09:46.630> projects<00:09:47.079> around<00:09:47.410> mobile 00:09:47.850 --> 00:09:47.860 align:start position:0% interesting projects around mobile 00:09:47.860 --> 00:09:50.670 align:start position:0% interesting projects around mobile devices<00:09:48.370> around<00:09:48.940> using<00:09:49.390> crowds<00:09:49.930> for<00:09:50.170> data 00:09:50.670 --> 00:09:50.680 align:start position:0% devices around using crowds for data 00:09:50.680 --> 00:09:52.710 align:start position:0% devices around using crowds for data labeling<00:09:51.130> and<00:09:51.250> so<00:09:51.399> on<00:09:51.579> that<00:09:52.029> that<00:09:52.360> started<00:09:52.630> to 00:09:52.710 --> 00:09:52.720 align:start position:0% labeling and so on that that started to 00:09:52.720 --> 00:09:54.449 align:start position:0% labeling and so on that that started to come<00:09:52.779> out<00:09:52.990> and<00:09:53.200> you<00:09:53.980> know<00:09:54.100> it<00:09:54.160> was<00:09:54.279> a<00:09:54.310> great 00:09:54.449 --> 00:09:54.459 align:start position:0% come out and you know it was a great 00:09:54.459 --> 00:09:56.639 align:start position:0% come out and you know it was a great line<00:09:54.670> of<00:09:54.790> research<00:09:54.910> so<00:09:55.870> this<00:09:56.290> is<00:09:56.410> kind<00:09:56.589> of 00:09:56.639 --> 00:09:56.649 align:start position:0% line of research so this is kind of 00:09:56.649 --> 00:10:00.240 align:start position:0% line of research so this is kind of humorous<00:09:57.839> semi<00:09:58.839> humorous<00:09:59.200> observation<00:09:59.290> about 00:10:00.240 --> 00:10:00.250 align:start position:0% humorous semi humorous observation about 00:10:00.250 --> 00:10:02.670 align:start position:0% humorous semi humorous observation about the<00:10:00.579> structure<00:10:00.940> of<00:10:01.120> all<00:10:01.329> employed<00:10:02.140> startups 00:10:02.670 --> 00:10:02.680 align:start position:0% the structure of all employed startups 00:10:02.680 --> 00:10:04.500 align:start position:0% the structure of all employed startups okay<00:10:03.010> and<00:10:03.100> if<00:10:03.160> you<00:10:03.190> go<00:10:03.370> to<00:10:03.760> their<00:10:03.970> you<00:10:04.300> know<00:10:04.420> if 00:10:04.500 --> 00:10:04.510 align:start position:0% okay and if you go to their you know if 00:10:04.510 --> 00:10:06.360 align:start position:0% okay and if you go to their you know if you<00:10:04.600> go<00:10:04.720> to<00:10:04.779> the<00:10:04.930> talks<00:10:05.140> at<00:10:05.800> some<00:10:05.950> point<00:10:06.160> they 00:10:06.360 --> 00:10:06.370 align:start position:0% you go to the talks at some point they 00:10:06.370 --> 00:10:08.220 align:start position:0% you go to the talks at some point they show<00:10:06.579> you<00:10:06.640> a<00:10:06.760> chart<00:10:07.060> of<00:10:07.329> the<00:10:07.420> architecture<00:10:07.899> and 00:10:08.220 --> 00:10:08.230 align:start position:0% show you a chart of the architecture and 00:10:08.230 --> 00:10:10.590 align:start position:0% show you a chart of the architecture and I<00:10:08.380> call<00:10:08.620> the<00:10:08.709> three<00:10:08.950> layer<00:10:09.130> and<00:10:09.490> black<00:10:09.670> pie<00:10:09.880> it 00:10:10.590 --> 00:10:10.600 align:start position:0% I call the three layer and black pie it 00:10:10.600 --> 00:10:13.500 align:start position:0% I call the three layer and black pie it basically<00:10:11.170> it<00:10:11.290> has<00:10:11.589> three<00:10:12.040> layers<00:10:12.450> at<00:10:13.450> the 00:10:13.500 --> 00:10:13.510 align:start position:0% basically it has three layers at the 00:10:13.510 --> 00:10:15.389 align:start position:0% basically it has three layers at the bottom<00:10:13.630> of<00:10:14.050> all<00:10:14.260> things<00:10:14.560> the<00:10:14.860> middle<00:10:15.279> layer 00:10:15.389 --> 00:10:15.399 align:start position:0% bottom of all things the middle layer 00:10:15.399 --> 00:10:16.530 align:start position:0% bottom of all things the middle layer reads<00:10:15.670> from<00:10:15.760> the<00:10:15.880> middle<00:10:16.180> layer<00:10:16.270> at<00:10:16.450> the 00:10:16.530 --> 00:10:16.540 align:start position:0% reads from the middle layer at the 00:10:16.540 --> 00:10:19.199 align:start position:0% reads from the middle layer at the startup<00:10:16.930> it<00:10:17.110> may<00:10:17.230> be<00:10:17.290> spark<00:10:17.649> and<00:10:18.310> I<00:10:18.399> be<00:10:18.610> a<00:10:18.940> Luxio 00:10:19.199 --> 00:10:19.209 align:start position:0% startup it may be spark and I be a Luxio 00:10:19.209 --> 00:10:21.960 align:start position:0% startup it may be spark and I be a Luxio or<00:10:19.899> the<00:10:20.079> mesosphere<00:10:20.560> all<00:10:21.370> right<00:10:21.520> so<00:10:21.670> it<00:10:21.820> sees 00:10:21.960 --> 00:10:21.970 align:start position:0% or the mesosphere all right so it sees 00:10:21.970 --> 00:10:23.639 align:start position:0% or the mesosphere all right so it sees on<00:10:22.089> top<00:10:22.240> of<00:10:22.329> stuff<00:10:22.510> so<00:10:22.779> tweets<00:10:23.110> from<00:10:23.410> are 00:10:23.639 --> 00:10:23.649 align:start position:0% on top of stuff so tweets from are 00:10:23.649 --> 00:10:26.579 align:start position:0% on top of stuff so tweets from are things<00:10:24.070> like<00:10:24.279> HDFS<00:10:24.880> or<00:10:25.089> s3<00:10:25.420> then<00:10:25.959> the<00:10:26.140> spark 00:10:26.579 --> 00:10:26.589 align:start position:0% things like HDFS or s3 then the spark 00:10:26.589 --> 00:10:27.900 align:start position:0% things like HDFS or s3 then the spark and<00:10:26.829> then<00:10:26.980> there<00:10:27.160> is<00:10:27.250> a<00:10:27.279> whole<00:10:27.459> bunch<00:10:27.670> of<00:10:27.760> apps 00:10:27.900 --> 00:10:27.910 align:start position:0% and then there is a whole bunch of apps 00:10:27.910 --> 00:10:30.389 align:start position:0% and then there is a whole bunch of apps like<00:10:28.240> in<00:10:28.510> my<00:10:28.600> lab<00:10:28.810> or<00:10:29.020> in<00:10:29.260> application<00:10:30.250> layer 00:10:30.389 --> 00:10:30.399 align:start position:0% like in my lab or in application layer 00:10:30.399 --> 00:10:31.920 align:start position:0% like in my lab or in application layer frameworks<00:10:30.850> which<00:10:31.060> is<00:10:31.209> on<00:10:31.360> top<00:10:31.540> of<00:10:31.660> the<00:10:31.750> pipe 00:10:31.920 --> 00:10:31.930 align:start position:0% frameworks which is on top of the pipe 00:10:31.930 --> 00:10:33.750 align:start position:0% frameworks which is on top of the pipe so<00:10:32.290> if<00:10:32.350> you<00:10:32.410> look<00:10:32.560> at<00:10:32.649> every<00:10:33.100> implant<00:10:33.490> you<00:10:33.640> know 00:10:33.750 --> 00:10:33.760 align:start position:0% so if you look at every implant you know 00:10:33.760 --> 00:10:37.170 align:start position:0% so if you look at every implant you know startup<00:10:34.480> deck<00:10:34.720> or<00:10:35.040> slide<00:10:36.040> presentation<00:10:36.550> later 00:10:37.170 --> 00:10:37.180 align:start position:0% startup deck or slide presentation later 00:10:37.180 --> 00:10:39.630 align:start position:0% startup deck or slide presentation later by<00:10:37.360> so<00:10:37.540> what<00:10:37.660> is<00:10:37.810> your<00:10:38.050> three<00:10:38.709> layer<00:10:38.980> part<00:10:39.310> yeah 00:10:39.630 --> 00:10:39.640 align:start position:0% by so what is your three layer part yeah 00:10:39.640 --> 00:10:41.460 align:start position:0% by so what is your three layer part yeah absolutely<00:10:39.910> so<00:10:40.329> sometimes<00:10:41.079> we<00:10:41.200> call<00:10:41.350> this 00:10:41.460 --> 00:10:41.470 align:start position:0% absolutely so sometimes we call this 00:10:41.470 --> 00:10:43.670 align:start position:0% absolutely so sometimes we call this lasagna<00:10:42.459> diagram<00:10:42.880> okay 00:10:43.670 --> 00:10:43.680 align:start position:0% lasagna diagram okay 00:10:43.680 --> 00:10:47.130 align:start position:0% lasagna diagram okay but<00:10:44.680> yeah<00:10:44.920> so<00:10:45.190> our<00:10:46.029> three<00:10:46.329> layer<00:10:46.720> pie<00:10:46.959> and 00:10:47.130 --> 00:10:47.140 align:start position:0% but yeah so our three layer pie and 00:10:47.140 --> 00:10:48.420 align:start position:0% but yeah so our three layer pie and you're<00:10:47.260> right<00:10:47.440> there<00:10:47.950> is<00:10:48.070> one<00:10:48.220> of<00:10:48.339> these 00:10:48.420 --> 00:10:48.430 align:start position:0% you're right there is one of these 00:10:48.430 --> 00:10:50.850 align:start position:0% you're right there is one of these things<00:10:48.640> and<00:10:48.880> I<00:10:49.300> was<00:10:49.390> talking<00:10:49.690> to<00:10:49.810> a<00:10:49.860> someone 00:10:50.850 --> 00:10:50.860 align:start position:0% things and I was talking to a someone 00:10:50.860 --> 00:10:52.650 align:start position:0% things and I was talking to a someone over<00:10:51.040> lunch<00:10:51.220> today<00:10:51.400> and<00:10:51.520> I<00:10:51.670> said<00:10:51.730> you<00:10:52.420> know<00:10:52.540> as 00:10:52.650 --> 00:10:52.660 align:start position:0% over lunch today and I said you know as 00:10:52.660 --> 00:10:54.370 align:start position:0% over lunch today and I said you know as computer<00:10:52.959> scientists<00:10:53.350> were<00:10:53.560> flagrant 00:10:54.370 --> 00:10:54.380 align:start position:0% computer scientists were flagrant 00:10:54.380 --> 00:10:56.110 align:start position:0% computer scientists were flagrant we<00:10:54.620> love<00:10:54.800> the<00:10:54.920> inventing<00:10:55.730> layers<00:10:55.910> of 00:10:56.110 --> 00:10:56.120 align:start position:0% we love the inventing layers of 00:10:56.120 --> 00:10:57.250 align:start position:0% we love the inventing layers of abstraction<00:10:56.270> that<00:10:56.630> everybody<00:10:57.050> else<00:10:57.200> should 00:10:57.250 --> 00:10:57.260 align:start position:0% abstraction that everybody else should 00:10:57.260 --> 00:10:58.840 align:start position:0% abstraction that everybody else should use<00:10:57.500> and<00:10:57.740> so<00:10:57.830> we're<00:10:57.980> guilty<00:10:58.130> as<00:10:58.340> charged 00:10:58.840 --> 00:10:58.850 align:start position:0% use and so we're guilty as charged 00:10:58.850 --> 00:11:02.410 align:start position:0% use and so we're guilty as charged but<00:10:59.750> our<00:11:00.290> familiar<00:11:00.890> PI<00:11:01.160> we<00:11:01.400> start<00:11:01.760> with<00:11:02.000> the 00:11:02.410 --> 00:11:02.420 align:start position:0% but our familiar PI we start with the 00:11:02.420 --> 00:11:04.660 align:start position:0% but our familiar PI we start with the under<00:11:02.900> layer<00:11:03.080> of<00:11:03.350> of<00:11:03.860> course<00:11:04.100> distributed 00:11:04.660 --> 00:11:04.670 align:start position:0% under layer of of course distributed 00:11:04.670 --> 00:11:07.240 align:start position:0% under layer of of course distributed storage<00:11:05.060> and<00:11:05.900> compute<00:11:06.620> framework<00:11:06.920> in<00:11:07.040> this 00:11:07.240 --> 00:11:07.250 align:start position:0% storage and compute framework in this 00:11:07.250 --> 00:11:09.790 align:start position:0% storage and compute framework in this case<00:11:07.460> mostly<00:11:07.970> we're<00:11:08.300> talking<00:11:08.570> about<00:11:08.660> GPUs<00:11:09.230> but 00:11:09.790 --> 00:11:09.800 align:start position:0% case mostly we're talking about GPUs but 00:11:09.800 --> 00:11:12.030 align:start position:0% case mostly we're talking about GPUs but also<00:11:10.340> you<00:11:10.910> know<00:11:11.000> things<00:11:11.330> like<00:11:11.420> edge<00:11:11.660> devices 00:11:12.030 --> 00:11:12.040 align:start position:0% also you know things like edge devices 00:11:12.040 --> 00:11:14.500 align:start position:0% also you know things like edge devices that<00:11:13.040> mobile<00:11:13.610> phones<00:11:13.790> and<00:11:13.940> so<00:11:14.060> on<00:11:14.090> that<00:11:14.360> might 00:11:14.500 --> 00:11:14.510 align:start position:0% that mobile phones and so on that might 00:11:14.510 --> 00:11:16.630 align:start position:0% that mobile phones and so on that might be<00:11:14.660> deployed<00:11:15.110> at<00:11:15.230> the<00:11:15.260> edge<00:11:15.380> for<00:11:15.530> inference<00:11:15.890> on 00:11:16.630 --> 00:11:16.640 align:start position:0% be deployed at the edge for inference on 00:11:16.640 --> 00:11:19.060 align:start position:0% be deployed at the edge for inference on top<00:11:17.000> of<00:11:17.150> that<00:11:17.270> sits<00:11:17.720> both<00:11:18.170> the<00:11:18.560> existing 00:11:19.060 --> 00:11:19.070 align:start position:0% top of that sits both the existing 00:11:19.070 --> 00:11:20.500 align:start position:0% top of that sits both the existing application<00:11:19.670> frameworks<00:11:20.060> for<00:11:20.210> deep<00:11:20.360> learning 00:11:20.500 --> 00:11:20.510 align:start position:0% application frameworks for deep learning 00:11:20.510 --> 00:11:21.550 align:start position:0% application frameworks for deep learning so<00:11:20.930> tensorflow 00:11:21.550 --> 00:11:21.560 align:start position:0% so tensorflow 00:11:21.560 --> 00:11:25.540 align:start position:0% so tensorflow cafe<00:11:22.040> Karis<00:11:22.460> PI<00:11:23.180> torch<00:11:23.420> you<00:11:23.570> name<00:11:23.750> it<00:11:24.310> we<00:11:25.310> tend 00:11:25.540 --> 00:11:25.550 align:start position:0% cafe Karis PI torch you name it we tend 00:11:25.550 --> 00:11:27.610 align:start position:0% cafe Karis PI torch you name it we tend to<00:11:25.610> be<00:11:25.760> we<00:11:26.060> we<00:11:26.420> try<00:11:26.570> to<00:11:26.630> be<00:11:26.750> as<00:11:26.840> framework<00:11:27.440> it's 00:11:27.610 --> 00:11:27.620 align:start position:0% to be we we try to be as framework it's 00:11:27.620 --> 00:11:29.290 align:start position:0% to be we we try to be as framework it's not<00:11:27.770> agnostic<00:11:28.370> as<00:11:28.490> possible<00:11:28.910> there<00:11:29.270> are 00:11:29.290 --> 00:11:29.300 align:start position:0% not agnostic as possible there are 00:11:29.300 --> 00:11:30.700 align:start position:0% not agnostic as possible there are dozens<00:11:29.540> of<00:11:29.750> these<00:11:29.930> frameworks<00:11:30.140> we<00:11:30.440> can't<00:11:30.620> do 00:11:30.700 --> 00:11:30.710 align:start position:0% dozens of these frameworks we can't do 00:11:30.710 --> 00:11:32.440 align:start position:0% dozens of these frameworks we can't do every<00:11:30.950> one<00:11:31.070> but<00:11:31.490> we<00:11:31.580> support<00:11:31.730> all<00:11:32.000> the<00:11:32.090> popular 00:11:32.440 --> 00:11:32.450 align:start position:0% every one but we support all the popular 00:11:32.450 --> 00:11:34.900 align:start position:0% every one but we support all the popular ones<00:11:32.600> that<00:11:32.780> we<00:11:32.870> can<00:11:33.050> okay<00:11:33.380> and<00:11:34.070> then<00:11:34.370> alongside 00:11:34.900 --> 00:11:34.910 align:start position:0% ones that we can okay and then alongside 00:11:34.910 --> 00:11:37.660 align:start position:0% ones that we can okay and then alongside that<00:11:35.390> we<00:11:35.630> have<00:11:35.960> our<00:11:36.350> stuff<00:11:36.710> and<00:11:37.010> and<00:11:37.280> above<00:11:37.520> it 00:11:37.660 --> 00:11:37.670 align:start position:0% that we have our stuff and and above it 00:11:37.670 --> 00:11:40.900 align:start position:0% that we have our stuff and and above it which<00:11:38.510> is<00:11:38.720> a<00:11:39.160> distributed<00:11:40.160> job<00:11:40.400> scheduler 00:11:40.900 --> 00:11:40.910 align:start position:0% which is a distributed job scheduler 00:11:40.910 --> 00:11:42.940 align:start position:0% which is a distributed job scheduler something<00:11:41.450> we<00:11:41.720> could<00:11:41.870> think<00:11:42.050> of<00:11:42.260> as<00:11:42.410> the<00:11:42.620> the 00:11:42.940 --> 00:11:42.950 align:start position:0% something we could think of as the the 00:11:42.950 --> 00:11:45.790 align:start position:0% something we could think of as the the trial<00:11:43.160> optimizer<00:11:43.670> a<00:11:44.380> hyper<00:11:45.380> parameter 00:11:45.790 --> 00:11:45.800 align:start position:0% trial optimizer a hyper parameter 00:11:45.800 --> 00:11:48.880 align:start position:0% trial optimizer a hyper parameter optimization<00:11:46.420> set<00:11:47.420> of<00:11:47.510> techniques<00:11:47.870> which<00:11:48.740> we 00:11:48.880 --> 00:11:48.890 align:start position:0% optimization set of techniques which we 00:11:48.890 --> 00:11:50.950 align:start position:0% optimization set of techniques which we call<00:11:49.310> the<00:11:49.430> experiment<00:11:49.880> manager<00:11:50.240> and<00:11:50.540> then<00:11:50.930> a 00:11:50.950 --> 00:11:50.960 align:start position:0% call the experiment manager and then a 00:11:50.960 --> 00:11:52.930 align:start position:0% call the experiment manager and then a metadata<00:11:51.650> story<00:11:51.950> that<00:11:52.070> really<00:11:52.190> traps<00:11:52.610> all<00:11:52.820> of 00:11:52.930 --> 00:11:52.940 align:start position:0% metadata story that really traps all of 00:11:52.940 --> 00:11:54.730 align:start position:0% metadata story that really traps all of the<00:11:53.030> things<00:11:53.240> you've<00:11:53.390> ever<00:11:53.450> tried<00:11:53.840> about<00:11:54.440> you 00:11:54.730 --> 00:11:54.740 align:start position:0% the things you've ever tried about you 00:11:54.740 --> 00:11:56.260 align:start position:0% the things you've ever tried about you know<00:11:54.830> developing<00:11:55.280> your<00:11:55.370> deep<00:11:55.850> learning<00:11:55.970> job 00:11:56.260 --> 00:11:56.270 align:start position:0% know developing your deep learning job 00:11:56.270 --> 00:11:59.410 align:start position:0% know developing your deep learning job over<00:11:56.510> time<00:11:56.710> and<00:11:57.710> then<00:11:58.160> the<00:11:58.880> front<00:11:59.120> end<00:11:59.270> is 00:11:59.410 --> 00:11:59.420 align:start position:0% over time and then the front end is 00:11:59.420 --> 00:12:00.940 align:start position:0% over time and then the front end is really<00:11:59.720> sort<00:12:00.020> of<00:12:00.080> command-line<00:12:00.590> interface 00:12:00.940 --> 00:12:00.950 align:start position:0% really sort of command-line interface 00:12:00.950 --> 00:12:02.350 align:start position:0% really sort of command-line interface set<00:12:01.130> of<00:12:01.220> API<00:12:01.460> is<00:12:01.580> to<00:12:01.790> integrate<00:12:02.120> with<00:12:02.210> the 00:12:02.350 --> 00:12:02.360 align:start position:0% set of API is to integrate with the 00:12:02.360 --> 00:12:04.600 align:start position:0% set of API is to integrate with the system<00:12:02.660> and<00:12:02.840> the<00:12:03.140> a<00:12:03.410> user<00:12:03.770> interface<00:12:04.070> to<00:12:04.580> help 00:12:04.600 --> 00:12:04.610 align:start position:0% system and the a user interface to help 00:12:04.610 --> 00:12:07.300 align:start position:0% system and the a user interface to help users<00:12:04.940> kind<00:12:05.270> of<00:12:05.330> understand<00:12:05.780> how<00:12:06.230> how<00:12:07.040> are<00:12:07.160> my 00:12:07.300 --> 00:12:07.310 align:start position:0% users kind of understand how how are my 00:12:07.310 --> 00:12:10.870 align:start position:0% users kind of understand how how are my models<00:12:07.490> doing<00:12:07.940> over<00:12:08.120> time<00:12:09.280> cool<00:12:10.280> cool 00:12:10.870 --> 00:12:10.880 align:start position:0% models doing over time cool cool 00:12:10.880 --> 00:12:12.700 align:start position:0% models doing over time cool cool so<00:12:11.000> it's<00:12:11.480> sort<00:12:11.660> of<00:12:11.690> sending<00:12:12.050> to<00:12:12.260> know<00:12:12.380> that<00:12:12.590> you 00:12:12.700 --> 00:12:12.710 align:start position:0% so it's sort of sending to know that you 00:12:12.710 --> 00:12:14.020 align:start position:0% so it's sort of sending to know that you guys<00:12:12.860> have<00:12:13.070> supported<00:12:13.460> multiple<00:12:13.820> frameworks 00:12:14.020 --> 00:12:14.030 align:start position:0% guys have supported multiple frameworks 00:12:14.030 --> 00:12:19.180 align:start position:0% guys have supported multiple frameworks and<00:12:14.390> so<00:12:14.600> that<00:12:14.780> that's<00:12:16.000> so<00:12:17.590> like<00:12:18.590> do<00:12:18.800> everything 00:12:19.180 --> 00:12:19.190 align:start position:0% and so that that's so like do everything 00:12:19.190 --> 00:12:21.610 align:start position:0% and so that that's so like do everything kind<00:12:19.280> of<00:12:19.480> pilots<00:12:20.480> in<00:12:20.660> the<00:12:20.780> works<00:12:20.960> can<00:12:21.260> you<00:12:21.290> tell 00:12:21.610 --> 00:12:21.620 align:start position:0% kind of pilots in the works can you tell 00:12:21.620 --> 00:12:23.260 align:start position:0% kind of pilots in the works can you tell us<00:12:21.650> a<00:12:21.770> little<00:12:21.800> bit<00:12:22.010> like<00:12:22.190> how's<00:12:22.400> it<00:12:22.550> going<00:12:22.850> yeah 00:12:23.260 --> 00:12:23.270 align:start position:0% us a little bit like how's it going yeah 00:12:23.270 --> 00:12:25.510 align:start position:0% us a little bit like how's it going yeah things<00:12:23.630> are<00:12:23.750> great<00:12:23.930> so<00:12:24.080> far<00:12:24.290> so<00:12:24.560> we<00:12:24.680> company's 00:12:25.510 --> 00:12:25.520 align:start position:0% things are great so far so we company's 00:12:25.520 --> 00:12:27.780 align:start position:0% things are great so far so we company's still<00:12:25.640> early<00:12:25.790> stage<00:12:26.180> we<00:12:26.510> we<00:12:27.290> were<00:12:27.440> founded 00:12:27.780 --> 00:12:27.790 align:start position:0% still early stage we we were founded 00:12:27.790 --> 00:12:30.880 align:start position:0% still early stage we we were founded late<00:12:28.790> last<00:12:28.970> spring<00:12:29.320> and<00:12:30.320> raised<00:12:30.740> some<00:12:30.860> money 00:12:30.880 --> 00:12:30.890 align:start position:0% late last spring and raised some money 00:12:30.890 --> 00:12:34.690 align:start position:0% late last spring and raised some money over<00:12:31.130> the<00:12:31.310> summer<00:12:31.540> we<00:12:33.010> spent<00:12:34.010> spend<00:12:34.490> a<00:12:34.520> bunch 00:12:34.690 --> 00:12:34.700 align:start position:0% over the summer we spent spend a bunch 00:12:34.700 --> 00:12:36.190 align:start position:0% over the summer we spent spend a bunch of<00:12:34.760> the<00:12:34.850> time<00:12:35.000> kind<00:12:35.450> of<00:12:35.510> building<00:12:35.780> on<00:12:35.840> product 00:12:36.190 --> 00:12:36.200 align:start position:0% of the time kind of building on product 00:12:36.200 --> 00:12:38.800 align:start position:0% of the time kind of building on product and<00:12:36.320> so<00:12:36.440> on<00:12:36.560> and<00:12:36.740> we've<00:12:37.160> got<00:12:38.000> a<00:12:38.450> few<00:12:38.600> paying 00:12:38.800 --> 00:12:38.810 align:start position:0% and so on and we've got a few paying 00:12:38.810 --> 00:12:40.860 align:start position:0% and so on and we've got a few paying pilot<00:12:39.140> customers<00:12:39.560> people<00:12:39.950> in<00:12:40.100> healthcare 00:12:40.860 --> 00:12:40.870 align:start position:0% pilot customers people in healthcare 00:12:40.870 --> 00:12:43.990 align:start position:0% pilot customers people in healthcare kind<00:12:41.870> of<00:12:41.960> the<00:12:42.230> genomic<00:12:42.620> space<00:12:42.710> as<00:12:43.070> well<00:12:43.220> as<00:12:43.280> as 00:12:43.990 --> 00:12:44.000 align:start position:0% kind of the genomic space as well as as 00:12:44.000 --> 00:12:46.090 align:start position:0% kind of the genomic space as well as as well<00:12:44.210> as<00:12:44.240> IOT<00:12:44.840> and<00:12:45.290> talking<00:12:45.590> to<00:12:45.650> some<00:12:45.770> folks<00:12:45.950> in 00:12:46.090 --> 00:12:46.100 align:start position:0% well as IOT and talking to some folks in 00:12:46.100 --> 00:12:48.430 align:start position:0% well as IOT and talking to some folks in autonomous<00:12:46.490> vehicles<00:12:46.940> and<00:12:47.180> and<00:12:47.840> financial 00:12:48.430 --> 00:12:48.440 align:start position:0% autonomous vehicles and and financial 00:12:48.440 --> 00:12:50.890 align:start position:0% autonomous vehicles and and financial services<00:12:48.800> as<00:12:48.980> well<00:12:49.100> so<00:12:49.430> so<00:12:50.150> far<00:12:50.300> so<00:12:50.420> good<00:12:50.720> we've 00:12:50.890 --> 00:12:50.900 align:start position:0% services as well so so far so good we've 00:12:50.900 --> 00:12:52.840 align:start position:0% services as well so so far so good we've got<00:12:50.930> a<00:12:51.080> few<00:12:51.290> real<00:12:51.500> users<00:12:51.830> kicking<00:12:52.490> the<00:12:52.580> system 00:12:52.840 --> 00:12:52.850 align:start position:0% got a few real users kicking the system 00:12:52.850 --> 00:12:55.030 align:start position:0% got a few real users kicking the system every<00:12:53.030> day<00:12:53.060> and<00:12:53.420> giving<00:12:54.230> us<00:12:54.380> feedback<00:12:54.410> and<00:12:54.950> so 00:12:55.030 --> 00:12:55.040 align:start position:0% every day and giving us feedback and so 00:12:55.040 --> 00:12:56.650 align:start position:0% every day and giving us feedback and so it's<00:12:55.160> been<00:12:55.280> good<00:12:55.400> to<00:12:55.550> see<00:12:55.580> very<00:12:56.210> cool 00:12:56.650 --> 00:12:56.660 align:start position:0% it's been good to see very cool 00:12:56.660 --> 00:12:58.509 align:start position:0% it's been good to see very cool so<00:12:56.960> in<00:12:57.170> those<00:12:57.350> missions 00:12:58.509 --> 00:12:58.519 align:start position:0% so in those missions 00:12:58.519 --> 00:13:01.009 align:start position:0% so in those missions what<00:12:59.519> we're<00:12:59.969> doing<00:13:00.179> to<00:13:00.269> be<00:13:00.299> the<00:13:00.449> year<00:13:00.629> where<00:13:00.959> as 00:13:01.009 --> 00:13:01.019 align:start position:0% what we're doing to be the year where as 00:13:01.019 --> 00:13:02.689 align:start position:0% what we're doing to be the year where as a<00:13:01.049> little<00:13:01.349> chicken<00:13:01.589> was<00:13:01.769> really<00:13:01.859> a<00:13:02.159> world<00:13:02.639> a 00:13:02.689 --> 00:13:02.699 align:start position:0% a little chicken was really a world a 00:13:02.699 --> 00:13:05.539 align:start position:0% a little chicken was really a world a world<00:13:03.089> domination<00:13:03.479> obvious<00:13:04.219> muscles<00:13:05.219> you 00:13:05.539 --> 00:13:05.549 align:start position:0% world domination obvious muscles you 00:13:05.549 --> 00:13:07.579 align:start position:0% world domination obvious muscles you know<00:13:05.639> I<00:13:05.669> think<00:13:05.849> that<00:13:06.209> there's<00:13:06.839> really<00:13:07.019> this 00:13:07.579 --> 00:13:07.589 align:start position:0% know I think that there's really this 00:13:07.589 --> 00:13:10.519 align:start position:0% know I think that there's really this opportunity<00:13:08.159> to<00:13:08.339> help<00:13:09.329> shape<00:13:09.599> the<00:13:10.079> landscape 00:13:10.519 --> 00:13:10.529 align:start position:0% opportunity to help shape the landscape 00:13:10.529 --> 00:13:11.749 align:start position:0% opportunity to help shape the landscape of<00:13:10.559> how<00:13:10.799> people<00:13:11.129> develop<00:13:11.279> deep<00:13:11.639> learning 00:13:11.749 --> 00:13:11.759 align:start position:0% of how people develop deep learning 00:13:11.759 --> 00:13:13.549 align:start position:0% of how people develop deep learning applications<00:13:12.479> and<00:13:12.989> you<00:13:13.079> have<00:13:13.229> these<00:13:13.379> great 00:13:13.549 --> 00:13:13.559 align:start position:0% applications and you have these great 00:13:13.559 --> 00:13:15.079 align:start position:0% applications and you have these great application<00:13:14.129> from<00:13:14.279> work<00:13:14.459> frameworks<00:13:14.939> the 00:13:15.079 --> 00:13:15.089 align:start position:0% application from work frameworks the 00:13:15.089 --> 00:13:16.789 align:start position:0% application from work frameworks the tensor<00:13:15.359> flows<00:13:15.569> the<00:13:15.899> carats<00:13:16.379> of<00:13:16.499> the<00:13:16.619> world 00:13:16.789 --> 00:13:16.799 align:start position:0% tensor flows the carats of the world 00:13:16.799 --> 00:13:18.859 align:start position:0% tensor flows the carats of the world that<00:13:17.369> are<00:13:17.489> really<00:13:17.519> good<00:13:18.059> it's<00:13:18.269> sort<00:13:18.479> of<00:13:18.539> model 00:13:18.859 --> 00:13:18.869 align:start position:0% that are really good it's sort of model 00:13:18.869 --> 00:13:21.409 align:start position:0% that are really good it's sort of model training<00:13:19.499> on<00:13:19.619> single<00:13:19.949> GPU<00:13:20.249> but<00:13:20.939> there<00:13:21.059> is<00:13:21.179> so 00:13:21.409 --> 00:13:21.419 align:start position:0% training on single GPU but there is so 00:13:21.419 --> 00:13:23.059 align:start position:0% training on single GPU but there is so much<00:13:21.599> more<00:13:21.899> to<00:13:22.109> the<00:13:22.199> application<00:13:22.709> development 00:13:23.059 --> 00:13:23.069 align:start position:0% much more to the application development 00:13:23.069 --> 00:13:25.669 align:start position:0% much more to the application development lifecycle<00:13:23.419> around<00:13:24.419> data<00:13:24.809> collection<00:13:24.899> around 00:13:25.669 --> 00:13:25.679 align:start position:0% lifecycle around data collection around 00:13:25.679 --> 00:13:27.499 align:start position:0% lifecycle around data collection around data<00:13:26.369> set<00:13:26.549> management<00:13:26.939> cluster<00:13:27.209> resource 00:13:27.499 --> 00:13:27.509 align:start position:0% data set management cluster resource 00:13:27.509 --> 00:13:29.629 align:start position:0% data set management cluster resource management<00:13:27.839> deployment<00:13:28.499> engineering<00:13:29.189> that 00:13:29.629 --> 00:13:29.639 align:start position:0% management deployment engineering that 00:13:29.639 --> 00:13:31.459 align:start position:0% management deployment engineering that needs<00:13:30.179> to<00:13:30.329> be<00:13:30.449> figured<00:13:30.629> out<00:13:30.839> and<00:13:31.079> there<00:13:31.199> isn't 00:13:31.459 --> 00:13:31.469 align:start position:0% needs to be figured out and there isn't 00:13:31.469 --> 00:13:33.769 align:start position:0% needs to be figured out and there isn't good<00:13:31.589> software<00:13:31.949> there<00:13:32.189> for<00:13:32.429> it<00:13:32.579> today<00:13:32.789> and<00:13:33.179> I 00:13:33.769 --> 00:13:33.779 align:start position:0% good software there for it today and I 00:13:33.779 --> 00:13:34.969 align:start position:0% good software there for it today and I hope<00:13:33.929> we<00:13:34.079> can<00:13:34.199> contribute<00:13:34.529> in<00:13:34.769> some<00:13:34.799> small 00:13:34.969 --> 00:13:34.979 align:start position:0% hope we can contribute in some small 00:13:34.979 --> 00:13:37.309 align:start position:0% hope we can contribute in some small part<00:13:35.489> to<00:13:35.519> that<00:13:35.699> and<00:13:35.939> and<00:13:36.239> make<00:13:37.019> the<00:13:37.109> world<00:13:37.289> a 00:13:37.309 --> 00:13:37.319 align:start position:0% part to that and and make the world a 00:13:37.319 --> 00:13:37.699 align:start position:0% part to that and and make the world a better<00:13:37.409> place 00:13:37.699 --> 00:13:37.709 align:start position:0% better place 00:13:37.709 --> 00:13:40.489 align:start position:0% better place any<00:13:38.329> wasteful<00:13:39.329> community<00:13:39.719> to<00:13:39.899> chip<00:13:40.109> and<00:13:40.259> then 00:13:40.489 --> 00:13:40.499 align:start position:0% any wasteful community to chip and then 00:13:40.499 --> 00:13:44.539 align:start position:0% any wasteful community to chip and then open<00:13:40.799> source<00:13:41.009> kind<00:13:41.609> of<00:13:41.699> points<00:13:42.539> so<00:13:42.869> so<00:13:43.199> far<00:13:43.549> no 00:13:44.539 --> 00:13:44.549 align:start position:0% open source kind of points so so far no 00:13:44.549 --> 00:13:46.819 align:start position:0% open source kind of points so so far no nothing<00:13:45.419> for<00:13:45.539> open<00:13:46.169> source<00:13:46.259> from<00:13:46.499> the<00:13:46.559> company 00:13:46.819 --> 00:13:46.829 align:start position:0% nothing for open source from the company 00:13:46.829 --> 00:13:48.109 align:start position:0% nothing for open source from the company we've<00:13:47.009> certainly<00:13:47.219> been<00:13:47.369> talking<00:13:47.789> about<00:13:47.879> which 00:13:48.109 --> 00:13:48.119 align:start position:0% we've certainly been talking about which 00:13:48.119 --> 00:13:49.789 align:start position:0% we've certainly been talking about which bits<00:13:48.329> of<00:13:48.479> this<00:13:48.629> should<00:13:48.989> be<00:13:49.019> things<00:13:49.649> that 00:13:49.789 --> 00:13:49.799 align:start position:0% bits of this should be things that 00:13:49.799 --> 00:13:52.279 align:start position:0% bits of this should be things that belong<00:13:50.009> to<00:13:50.129> the<00:13:50.219> community<00:13:50.659> we<00:13:51.659> we<00:13:52.019> haven't 00:13:52.279 --> 00:13:52.289 align:start position:0% belong to the community we we haven't 00:13:52.289 --> 00:13:53.829 align:start position:0% belong to the community we we haven't announced<00:13:52.559> a<00:13:52.739> formal<00:13:53.009> strategy<00:13:53.189> there<00:13:53.579> yet 00:13:53.829 --> 00:13:53.839 align:start position:0% announced a formal strategy there yet 00:13:53.839 --> 00:13:56.119 align:start position:0% announced a formal strategy there yet certainly<00:13:54.839> you<00:13:55.049> know<00:13:55.169> working<00:13:55.559> on<00:13:55.679> making<00:13:55.889> the 00:13:56.119 --> 00:13:56.129 align:start position:0% certainly you know working on making the 00:13:56.129 --> 00:13:58.309 align:start position:0% certainly you know working on making the the<00:13:56.159> application<00:13:56.909> frameworks<00:13:57.269> better<00:13:57.479> is<00:13:57.989> a 00:13:58.309 --> 00:13:58.319 align:start position:0% the application frameworks better is a 00:13:58.319 --> 00:14:00.049 align:start position:0% the application frameworks better is a is<00:13:58.739> a<00:13:58.829> big<00:13:59.069> way<00:13:59.189> that<00:13:59.399> that<00:13:59.669> people<00:13:59.939> can 00:14:00.049 --> 00:14:00.059 align:start position:0% is a big way that that people can 00:14:00.059 --> 00:14:02.719 align:start position:0% is a big way that that people can contribute<00:14:00.329> so<00:14:01.229> I'll<00:14:01.319> great<00:14:01.529> well<00:14:02.129> I<00:14:02.369> wish<00:14:02.579> you 00:14:02.719 --> 00:14:02.729 align:start position:0% contribute so I'll great well I wish you 00:14:02.729 --> 00:14:04.039 align:start position:0% contribute so I'll great well I wish you good<00:14:02.879> luck<00:14:03.029> in<00:14:03.089> looking<00:14:03.599> for<00:14:03.720> as<00:14:03.809> you<00:14:03.869> talk 00:14:04.039 --> 00:14:04.049 align:start position:0% good luck in looking for as you talk 00:14:04.049 --> 00:14:05.449 align:start position:0% good luck in looking for as you talk great<00:14:04.559> thank<00:14:04.859> you<00:14:04.889> very<00:14:04.949> much<00:14:05.220> appreciate 00:14:05.449 --> 00:14:05.459 align:start position:0% great thank you very much appreciate 00:14:05.459 --> 00:14:09.900 align:start position:0% great thank you very much appreciate your<00:14:05.759> time<00:14:05.819> thanks 00:14:09.900 --> 00:14:09.910 align:start position:0% 00:14:09.910 --> 00:14:15.150 align:start position:0% you 00:14:15.150 --> 00:14:15.160 align:start position:0% 00:14:15.160 --> 00:14:18.370 align:start position:0% [Music]