Launchpad Studio with Malika Cantor and Peter Norvig: GCPPodcast 108

Launchpad Studio with Malika Cantor and Peter Norvig: GCPPodcast 108

Show Video

Hi. And, welcome to episode number 108. The weekly Google cloud platform podcast. I'm mock mando and I'm here with my colleague Melanie work how, are you doing today Melanie Happy New Year mark it's the beginning of the year trying, to get back into the groove of things trying to. Did. You have a good holiday I did how was yours it, was good it was good it was low-key which is nice nice so this, week we've, got some friends of yours yeah. They're my friends your hands we hang out all the time I figure out the pool yeah yeah so, yes this week we are going to be talking to Malika Cantor, and Peter Norvig and they're going to tell us all about launchpad. Studio which is basically, a program that supports startups. Especially machine. Learning startups, before. We get into the interview we're, gonna of course start, off with the cool things of the week and at, the end with the question of the week and the question of the week is going to be about how if you're a startup, how would you get credits to work on GCP before, we get into the cool thing of the week what we do want to do is briefly mention, about vulnerabilities. That have come out last, week around on modern computers and specifically, these vulnerabilities, are related to modern processors, and they. Impact, things like your personal computer your, mobile devices and the cloud now, there's, a lot of content out there there's many. Different links, and will definitely provide some of these on our web site and, we have some, great information that, our security, group has done on a blog. That we will provide, around GCP and how, they're working to support, that, but. The main things to understand, is that meltdown inspector, are two vulnerabilities, that you want to be aware of you, want to make sure you're updating your operating system you want to make sure that you are updating, all the things whenever you have the opportunity, to on your phones and your laptops on, whatever cloud services you're working with and Mark, I think you wanted to touch on this briefly as well yeah we're gonna link to all the blog posts we are not security experts I think Melanie knows probably more about this than I do she can talk about freitas but most, importantly, yeah definitely upgrade all the things one, of the things that I'm just gonna literally, read off here for like Google cloud right so if you're writing compute engine kubernetes, engine, data flow data proc they're, awesome two extra, customer, actions that are needed again, links are in the in the show notes but definitely, go through those make sure you get those implemented, as soon as you possibly can.

So Check, out the links make, sure you're up to speed on that and let's. Talk about cool things of the week yeah absolutely so, every, time we seem to come on the podcast we talked about GPUs getting cheaper so. Why, don't we talk about GPUs getting cheaper so we're introducing, preemptable GPUs, which, is kind of neat so if you ever used preemptable, of GCE instances, so. Instances, that may, die and put about 24, hours but you get a very large deduction, in price when, you use them you're, now able to use GPU. Instances, on those now beta unavailable. For GPUs, on preemptable, instances we'll have a link in the show notes as well to the blog post which links to the documentation and how you can do them but basically if you wanted to batch. Type processing, for GPUs and you want it to be a lot cheaper to do it especially if you're using printable instances already this, is a really nice thing for you yes cheaper, is always better all, right and the, other cool thing of the week is that, we have a couple of awesome lists, that we wanted to share with everybody we've got some links we'll put it into the website as always but, specifically, if you've played around with github but you've seen the awesome and awesome list of Awesomeness, there. Are some, awesome lists, for tensorflow they're awesome list for kubernetes, there's, awesome lists for GCP, Google, cloud platform itself, and the awesome lists are really you know using contributors, contributing. What they think are some, really great training, resources, as well as real, world application, samples, libraries. Analytics. They put out there what they think, are good tools, and resources to have access to and consolidate. That down into a list and, I think mark you said you found in here something that you guys had built yeah I didn't, I didn't realize that I was just looking at this now there was a gr PC sample that myself Sandeep Dinesh, and a couple other developer advocates also did, some work on where we took the old game Simon with the multi colors and made it playable over the Internet we're, using to your PC in communities and that's on the list which. Cool. And finally, is, a really nice cool thing of the week that is coming from the community which, is really cool I'm, gonna butcher your name I'm really sorry I'm, gonna will Fuchs maybe, a, data, scientist, at defeated, group wrote, a medium post called how we implemented, a fully serviced, recommender, system using GCP, is a, really cool article in they use a pen big query could cut storage data product data store data flow to, build a recommendation engine but they also talk through the recommendation, algorithm, which, I love is called dim sum I don't know why thanks les that's fantastic, basically. To build a collaborative filtering recommendation. Engine which is yeah, it's really good read you should totally check it out thanks, all, right well now that we know what our cool things are the week are we're gonna go ahead and jump into the main content so we. Are excited. That we have with us malika. Cantor who is the global lead, of launchpad, studio and peter norvig who is the research, director of, machine, learning at Google so, before, we get started to, talking about launchpad, studio we, want to talk about a little bit about your background so please that reduce yourself a little more yes. So my name is Malika Kanter as, you mentioned I'm the global lead at launchpad, studio, launchpad. Sits, under, developer, relations at, Google we, are basically the accelerator, engine of Google so we run different, accelerator. Initiatives, focused on supporting, various startup, ecosystems, across the world and studio. Is focused on supporting specifically, applied AI and machine learning companies, globally, Thanks, and Peter, we don't really know who you are so you mind telling us a little about yourself. Hi. I'm Peter Norvig I'm a director, of research here at Google I've been at Google since 2001. Company, was a little bit smaller than and. Been. Focusing, on, machine. Learning for a long. Time working. On various. Tools and applications and. About. A year ago started getting involved with launchpad. To help. Others learn, how to use those tools and. You did write the canonical, book, on AI that, most people tend to look, back to was Stuart Russell that's, right, so you.

Know I'm in the comm. Domain, now but I've had all the top levels so I was in Eadie, you was professor for a while I, was in Gov, in computer, science at NASA, and along. The way I, wrote. A textbook and kind of got involved right, about the front a little bit of background there, I know Mullica we were talking about it too that you've had some great experiences around, the VC space as well as especially, in China and we want to get into that a little further later but first let's, talk a little bit about launchpad. Studio so you're. Explaining an, overview, of it but what. Led to this project what, inspired, you guys to to get going with it sure, so I, mean as you mentioned I was, working in venture capital, investing. In applied AI machine, learning and robotics companies for a little while and one. Of the things I realized was, that a lot of the startups didn't, really have the tools that they needed to be successful and so, one of my thoughts was well where can I go and what can I do to actually make a difference and, this. Was about six months ago I started like, Peter getting involved with a launchpad team and saw, a great opportunity to, build a program that would actually, allow us to gather more insights on what startups. Working. On applied machine, learning in different industries, actually, needed and for us to then go back to Google, and other. Potential, large tech companies and inform. Them on what were the tools that these types needed to actually implement. Things at scale and grow and and be, successful so, that, was sort of why I got, involved and very. Early on you, know we started, talking with Peter about his, interest, in you. Know again, gathering these insights, almost. Like developing, a lab, where, we were working, with great companies, you. Know trying different things with them bringing them amazing. Mentors and. Then figuring out if there were any sort. Of learnings, across these different companies and industries, that would then indicate, that there was you know that an opportunity, to build a tool on the, Google side and so this could be an API this could be a new type of service, that. Would then have, impact, across. The startup ecosystem and. Everywhere, in the world and across industries and the program, itself it's a six-month incubator. Program, to, bring in startups. Is that right effectively, yes so it's a six month program the. Way so once. We decided that we basically wanted to develop all of these insights we started thinking sort of from first principles, how it would make sense to set it up and we, decided together to focus on one specific industry. Vertical, at a time to. Allow us to really go deep and understand, what was going on there and and really. Gathered, really solid insight so right now we're focused on the applications, of AI, machine learning in healthcare and biotech, so. We, basically. Onboard, companies on a rolling basis, that are focused, on this vertical we, already have four companies, that we've been working with for a couple of months and. Checking. In with them every couple of weeks they come in and work with us every few months and we'll be onboarding, new companies, actually just in two weeks we. Have a big next, event and so, they come in they, meet with some a lot of different product teams at Google but also mentors, outside, and then, in this whole process we're, basically gathering. Information about, some, of their challenges. So for instance you know I've been using tensorflow but I just can't figure out how to you. Know scale, up this part of my infrastructure, or you, know I have this data set but you know especially, in the healthcare space you know the disease I'm looking at only allows, for me to have quite a limited data set how can i amplify and, augment, that data set to actually use some of the models that are best, used, in the machine learning space a lot of these types of questions were already starting to see and. I'd love to talk more about some of the other sort of overarching challenges, that were already starting, to identify so, you're actually talking about that at those you said in the beginning you said there, are these tools that are missing as those the tools that are missing for those sort of problems that people are trying to solve is that is that kind of the gist yeah and I might also let Peter jump in here I think where all of this is very much a hypothesis.

You Know we sort of came at this and and maybe a good analogy, is you know we were thinking all, right you know Lean Startup methodologies. Like we've, been accustomed. To hearing this term and we were thinking about it with machine learning and this is something I started seeing already and in my, VC days is does it really work like can you fail fast can, you pivot as a company. That's leveraging machine, learning and our. General, sort, of you know understanding. Was well no actually we don't really think that it's possible, and it's, because, the tools are lacking like there are some tools like you know tensorflow is incredible, tool, GCP. With we have a lot of other ones out there and you know other people are building great tools too but it's, still extremely difficult, as a company. Leveraging, these new technologies, to do. What we think startups, should be able to do and so, through studio our hope is to really try to understand. What, are the limitations, of the tools that are already on the market which are quite limited to be honest and then inform, what are tools that we should be building to hopefully, make it easier and easier for startups to pivot and to. Fail fast in the machine learning space. So. I think it's a combination of, the tools and, best. Practices, and, the knowledge, and experience. So. I came to this you, know I've been working internally. With, teams who, they're you, know a large percentage, of the team of, PhDs. In machine learning right. And so they, had some background, and intuition, about what. To build but, it was still really hard and now. We want to say well, not everybody has that luxury of being. Able to have. A bunch of experts, of machine learning on your team maybe you're you're. In your domain whether it's healthcare. Or energy or transportation or. Whatever your your your, company is doing and. You mostly have experts, in that and they don't have the, background that, we had access to but. You have an idea that you have some data you want to do something useful with that data and you want to produce a product how can we do that and some of it is having the right tool but, some of it is just knowing what the whole process is of how you go about starting. From the raw data or even how, do you go about collecting data if you don't have any yet and, building. The whole pipeline. And evolving. That and keeping going and we. Just don't have experience with that we have we've, done a lot of one, by ones but. You know if you look at history. Of software engineering, we've. Been figuring out how to build software for 50 years and, we've only been trying to figure out how to build, machine learning system for a, small, number of years so so. Understanding. How all that works is the key challenge now and. You've had the resources finally, to be able to experiment with building, up this machine learning but barriers. I know or definitely yeah, so, you see you, know there has been this big explosion in, the last couple of years and and I think that's a couple, of things coming together a lot of it as you say has to do with the resources that now. More. And more data is online, that is you know the internet is everywhere, and data is being captured in digital form so. That's available far. Greater computing, power that wasn't available in the past be able to do things that you couldn't try before and, just I. Think one success builds on another that, if somebody proves they can do something and then somebody else says uh well. That, work for them I think I can do it now too one. Of the four companies that you're working with currently what, are some of the tools that they are using and. Then also some of the tooling, that you both think could, definitely I mean we're, talking on it but anything specific, that's standing out that you think could be useful, yeah.

So, In, the companies right now again, we're like in the early days of the, program a lot of it is you, know mentorship. You, know giving. Insights, in in, terms of how they could be leveraging for instance GCP better some of the api's that we've already built you. Know they're working with teams like the cloud machine learning engine team you, know the cloudy eye team but, then also the Android things team which is trying to sort, of push a lot of learning onto the device now you know we're basically working with a lot of different product teams are coming in and mentoring startups. On what, are some of the services that we have already that some of these companies could be leveraging, as part of their stack so that they don't build that themselves because, there's there's, no point there you, know what's interesting, to. The software engineering, point that that Peter was talking about is you know a lot of the the issues that we're seeing the start are having is actually around architecture. And around you, know with software, development you're thinking about building an end product but with machine learning you're, thinking about how, to build a scalable data, pipeline, and how to retrain, a model and it's you, know the end product, is almost like a, byproduct. It's it's, not necessarily, what you're moving towards and so we're. Starting to try to figure out you know are there things that we could be building and we definitely are on the way to this already with initiatives. Like Auto ml in terms of you know automating. Certain parts of you know this. This process, of building, you, know your, software, engineering and development. To, help these startups, do this more quickly and, better you know the concept of potentially, automating, DevOps, also, like I think there's a lot of opportunities, here but. I think that that is one thing that we've seen that's been really interesting sort of aside. From other, problems, like I mentioned around you know data UI, UX there's, also you, know it's, a Peter's point there. There aren't tools for understanding how. Regulation is, going to affect your product, and so, there it becomes more about best practices, and sharing best practices, across these different companies that are working in one industry you, know our hope is maybe that we can even build a curriculum to try to help, inform, you know okay well be aware of regulation.

Be Aware of how, to go about building you. Know how to do software development how, to think, about your data pipeline, how to think about UI and UX and so again. To Peter's point it goes beyond just the idea of tools it's. More about sort of sharing, best practices, and learning from one another so, it's actually sounds really interesting I'm quite, curious you're talking Peter about sort of the, scarcity of resources in terms of people people with the knowledge are, you finding, that one of the big challenges is that people who may come in without that machine learning Rai knowledge, have, anymore I'm gonna say quote unquote traditional. Software development backgrounds so the pipeline and that the way that they build things for machine learning in a eyes is I don't know how different but different, enough that it's causing a disconnect, or that's like a big hurdle you finding that a challenge yeah, I think that's right, so. So. One of it is really, changing the mindset for what you're trying to do right. Traditional. Software, engineering, you're, kind of thinking like a mathematician. Or a logician and you're, you. Know you say well in the end point I could, sort, of prove my program correct, and you. Know in reality you only do that in the classroom you don't do it for any realistic program but. You're sort. Of aiming in that direction I mean we have tools like, test. Frameworks, where. You're making assertions and, you're the assertions are of the form. The. Result of this should be equal to that or, this should be true, so. It's very binary. True/false. Either, you got it or you don't. But with machine learning it's not like that it's not mathematical. It's. More like a, Pyrrhic, allure scientific. Endeavor, I was like doing biology where, you do an experiment on the world you. Observe the results of the experiment, it's not that it's true or false it's just you, know most of the time is going to be this and some of the time it's going to be something else and so. Making that change from thinking from. Logic. True/false. To a kind. Of probabilistic. Messy. World. That's. Hard for a lot of people to give up on on what, they thought was the solid bedrock, and, now, embrace, this uncertainty, kind, of via what do you mean this is only going to happen sometimes, yeah so one of the things I wanted to know is what do you think is making your, or what kind of companies, do you accept. Into the program what makes them successful in, this type of program, yes. So our selection. Criteria. Is. Quite simple so first. It's are you applying machine learning or ai to the, industry, that we've, said, we're focused on at the present time which is healthcare, and biotech right now and we're going to be announcing other industry, tracks in the course of 2018. The. Next one is are, you, trying to solve a. Specific. Challenge related to machine learning in the next six months and. The reason why that is part of our selection criteria is, we don't want to, basically. Take. Startups, in the wrong direction like, there are companies, that are going, to be implementing, AI machine, learning at some point there are some companies that never will and we, just want to make sure that the companies that we do work with because, we put a tremendous amount of resources and time on our end but we also expect, a lot from the companies in terms of being, there in the program, you, know, implementing. A lot of the the feedback that we give them we, want it to be core to what their, roadmap is in the next six months one, thing that I should clarify is our main Engagement. Mechanism, with the startup so we don't actually invest in the companies it's.

A Slightly. Untraditional. Accelerator, program, in that we, actually get them to draw. Up a project proposal of, what they want to accomplish with, Google, in the space of six months and we, select, them based on their project proposal, so they actually have to give us metrics, you. Know deliverables. Goals, for, what they want to achieve. And then we give them a lot of feedback, in the selection process and then based on that project we actually admit. Them into the program what kind of metrics, it. Really, would depend so you know if you're a computer vision company looking, at cell dynamics, and, your project, is focused on improving, precision. Accuracy of detection of a certain kind of disease looking, at those cells then it would be improving. Your precision accuracy from 83% to 96% and that, would be the metric that we're basically working, on, together but if it's a company, working the natural language space then you know it might be slightly, different it might be more about you know what are how many errors can, we detect. In. Speech, it, could be so it really it depends from one company to another but. Yeah that's I'd, say is is the main selection. Criteria, for sure for, us we you, know it's it's less about, selling. Google products, or you know it we don't really have a very clear. Sort of direct. You. Know gaining, back to Google but Peter, and I definitely when, we're looking at these companies are thinking about what can we learn and. Sometimes. It's even more just from a personal, intellectual. Perspective, of wow this company is doing something that we never thought would be possible and we think there's something really. Interesting to learn here but, definitely were as, I mentioned earlier you, know trying, to figure out what, Google could be doing and building to make this whole ecosystem, successful, so the, more company gives us insight. Into that the the better it's going to be because we're trying to do things at scale here beyond, just making, these for companies, and and other companies in the program successful, too so. I'm actually curious almost at a higher level to. You. You you obviously verticalized. It in the ml space but now I'm also curious like why. Specifically. Focusing on ml like the set up space is huge we do a lot of stuff across Google, cloud why. The emphasis is specifically on M oh yeah. And, you know I'll definitely let Peter, jump in here too, from. A launchpad perspective. So there's actually an another phenomenal.

Program That's being run out of launchpad called Launchpad accelerator, that's been running for. A few, years now been extremely successful and they what they focus on is supporting, growth stage startups and emerging markets and the whole vision of launchpad and so, we're affiliated with ecosystem, which you know our whole vision. And mission is to support startup, ecosystems, everywhere in the world is to, identify. Startup. Ecosystems, that we think are particularly, under. Supported, and so, with accelerator, it was you know sort of thinking about next billion users what. Could we be doing with startups, globally, that would actually, give us insights, and really make an impact and then, with studio again the thought process was similar is you know we have this as Peter, was saying this revolution. Going on this boom of like, more and more startups, that are starting to you, know for, better or for worse use. AI machine learning in certain cases not, in a very real way but, you know there. Was we identified a, need and, an also an opportunity for Google, to really make a difference here you, know to as, Peter said use, some of our expertise, of building, some of these products internally, and actually. See if we can teach anything and learn anything from these companies are going really deep looking. At specific use cases and trying, to scale and, productize some of these, things we've been working on for a few years yeah. You know I think at, Google we have the luxury of in. Some ways being able to see into the future that, were, at such, a scale that we run into some, of the problems and opportunities. Before. Other, people who had, a smaller scale so we've been doing this for a while and we've seen, these opportunities. And now we see a way to share. That to to, help other companies, overcome. What, we've done in the past that they're just starting to face now, you. Were saying before that there's mentorship, that's provided, it's a six months program. Is it similar to some of the other incubators out there or is it different is there certain things that make it similar or different yes. So I think as I mentioned earlier, the main difference, is that we don't actually invest, we don't have a financial relationship with. The, startups that we work with there's, various, reasons why, that, is the case you. Know one, of them being that actually it would be very complicated for, Google to have a financial. Relationship with, some of the startups that we're bringing in from different places in the world you. Know you mentioned China earlier. It's. You, know we're very excited to be able to work with companies without. Having, some of that red tape but, it's also because, we want to make it very clear that this is not a strategic, endeavor for, Google and that this is you, know almost a room of peers, of you know we, have a lot to learn from you we, you, know in many cases we have mentors, from our product teams at Google who you know come to us can tell us how much they've learned from the stirrups that they've been working with so. That's one main difference the, other one is as, I also mentioned we admit, companies on a rolling basis, so we don't really believe in the concept of classes, you know from from my background working with startups and peter has to a lot of people on our team were very conscientious. Of the fact that you, know startups can't just drop everything you, know every like, once in a while and and and say okay we're gonna go in and join, this program and, and you, know basically not, focus on product development for however long and and so the entire program, is designed in a way that's trying to be very mindful of that and is, saying look you know you're, closing, your round right now it might not make sense for you to join this program but in a month when we onboard our next batch like we'd love for you to to.

Come In so, that's the, other one the, third aspect I'd say that it is a little, bit different is that we have no size fits all one, size fits all so it's. Um it's completely, tailored to each company we. Admit companies, that have 12. People and have raised you know $100,000. And companies, that have about thousand people and raised seventy million dollars so that's kind, of the spread that we already have in our in our current batch and, so you can imagine that a company that has 12 people and companies as a thousand, needs very different kinds of support so, we actually really focus on you, know identifying. What are the challenges of each company and then build a totally. Custom sort. Of mentorship program, and and solution. For for each company which is one of the reasons we only work with very few companies at a time, cool. You talked a lot of you talked about sort of that feedback loop which sounds fantastic and I think that sounds wonderful I'm super curious to know what's. Been the most sort of interesting, or surprising or, like, what, feedback, that you've got or learning, that you've had from working with these companies. Peter. Do you have any so far that you'd like to share. I, think, it's just that, companies. Are really focused on. Solving. Their problem, and serving, their customers and, so they see the, technology. As a, way towards that goal and. The. Technology. Itself exactly what it is they. Don't care as, long as long, as you can get there and so. Making. It easier for them to take, advantage of that is, what. We're trying to do so one of the questions that I had you. Both were saying this and I've heard this a lot lately where people will interchange AI and m/l how, do you define. It how do you think, about it in regards to especially, working with these startups. So. I guess i i defined, artificial. Intelligence, as doing. Something smart so. Figuring. Out what you want to do and and doing it well in a way that, you didn't. Know how to do before you, know it so, in some ways i say that. Regular. Software is. Defining. Program. To do something that you know how to do and AI. Is doing, it for something that you don't know how to do. And. Then, machine. Learning just means you're, going to do that by learning. From examples rather. Than having. The programmer write down some rules or something. And, back. In the 1980s. We. Had the term AI. And. We also have the term expert, systems and what. That meant is you got an expert in, medicine. Or whatever it was and you interviewed them and you figured out what they knew and then you wrote. Down rules. That represented, what they know, and. So that was AI. By. Applying, expert, knowledge figuring. It out by hand now. We tend not to do that we find that that's brittle and slow and instead, we do it by collecting, data and learning from the data so machine learning it's achieving, AI through. Applying examples, and, expert. Systems is achieving, AI through. Understanding. Knowledge and. We also were talking a little bit before I just wanted to touch on it briefly before we start to wrap up but you've, got this great experience working in China and seeing, kind of where their world is going around the AI space so just as a general thought of where you, know AI is leading, us for for applied in real.

World Business applications. What's, your perception of this ecosystem. Yeah. I think it was extremely exciting to see Feifei, open. The AI Research, Institute, Beijing, you know mid, December last year a lot, of interest. And excitement, and, a lot of people working on it and, we're. Seeing new things coming out so. For, anyone who's listening and, is all like I've, got a business and we're doing ml and this program sounds amazing how. Do they get involved is there a way yes. So just. Go on our website there's an application, it. Only takes about ten minutes to fill out, we, would love to. Get. An application from, you you're. Actually, allowed. To apply even if you're not working on healthcare and biotech because, we're, looking, at applications, to also, influence. The next industry, verticals that we'll be announcing so. Again. We're, admitting. Companies, on a rolling basis. Applications. Are open across, industry. Verticals, but. We will be working mostly with companies, in the healthcare and biotech, space in the next few months also. If you aren't, a startup but would like to refer company. Please reach out to me and maybe, we can even share my information yes, well we'll definitely have them the show notes and, then the final thing is I'm, also if you are a subject-matter, expert, because I think one of the big themes in in this talk was basically. Effectively. Moving AI machine, learning out of the hands of just AI experts, and moving them increasingly, into the hands of subject matter experts like healthcare, and biotech experts, if you are a healthcare and biotech expert or someone, else in another industry you're, really passionate about, applying. Machine. Learning and AI to your industry and you'd like to become a mentor, again, please reach out we work with academics. With you, know industry, veterans, with, X CEOs. We'd, love to hear about your. Experience. And and you. Know your expertise, and nice. That, you give, it such a breadth in terms of being willing to take in the mentors in, terms of, one. Of the things I know we were talking about before was that you guys have an event coming up in a couple of weeks do, you want to tell us a little bit about that sure so basically it's. Our second. High touch point so that's what we call the. Event when, the startups, basically fly in in certain cases we have two companies that are based outside the US and two companies in the u.s. right now and they, come and basically, get a few days to a week of, back-to-back. Workshops. And meetings with a lot of Google teams and external, mentors like I mentioned so our. Next high touch point is is coming up we're going to be onboarding, a few new companies, so. You, know if you're interested, in seeing what, other types of use cases and companies we're gonna be supporting moving forward definitely keep an eye on our website, and on our Google. Developers, blog, and. Yeah, we're really excited to to keep working these companies and as, we mentioned you develop, product, methodologies, from machine learning and hopefully build, better tools for this whole ecosystem, thank you is there anything else that you both wanted to touch on before we close. Out this is really great, to hear about lunch pod studio and get, some insights in terms of how you're supporting startups, yeah. I guess the only thing is if, peter has any other thoughts because you know him. And I have had a lot of conversations, about this about you know his interest, and passion and again, you know basically. Empowering, people who aren't yeah experts, across. Different industries like, I guess we. Really enjoy working with companies and understanding. What their needs are and and. What kind of tools they, want and and, and. How we're gonna build that and. We. See that as real, challenges, going forward you, know so what I think a good example of that it's with speech, recognition, where. We've made great strides in, the last few years where, now you. Can talk to your phone and we're gonna get most of the words right. But. Now there's a next big step of you got all the words right but how do you respond to those words and. And. That, doesn't have to do just with language, that has to do with what is it that you're trying to achieve and, so we've got to give the companies the, capability. To say when. I recognize these words here's. What I should do and, do. That in a way that. They. Can understand, and build a good product without having, them having to be experts, in speech recognition and. Machine learning and, linguistics, and everything else but. Just integrated, into their product and, you. Know again it's a it's like in regular, software engineering we figured out how to build menus. And buttons and, mouse clicks, over. Decades and, now we have to do the same thing with a with a speech interface, and.

It's All going to be no and and. We don't know how to do it yet so we're we're looking to partner, with companies and help figure it out together and. Say thing to Peter's points almost like come, and help us learn you, know and and teach us what you need and you. Know we're really excited about working, with a lot of people so and. Actually just one another question here what got you inspired to do AI in general like what got you into the field if I may ask slight, side note. It. Just, seemed like a really exciting field and you. Know I was in high school and, I was lucky. Enough that my high school had both a computer. Class and a linguistics, class so it was rare that you, use had. Computers in the high schools in those days and and. I took those two classes and said hey could we put these two together and, my. Teacher wisely. Said yeah, it's possible but not with what you know so ever. Since I've been trying to learn, enough to be able to do it that's. Great that she gave you that kind of challenge, and that you took the challenge I'm, glad you both were able to join us thank you so much for sharing the information around, Launchpad. Studios and. Anything. Else that you wanted to mention or anything else you wanted to plug before we wrap this up all. Right well thank you very much to both of you for taking the time today and having a chat with us yeah thank you so much for having us you know thanks thanks, Monica and Peter for that great interview. We appreciated, the insights around launchpad, studio especially for, those startups, out there that are machine learning and want to have access, to different tooling, and and resources. If, you have any questions or want additional insights on launchpad studio we will as always put links on our website and, now. Let's, talk about the question of the week so, mark if, I am a startup and I want credits to work on TCP what is available to me yes, so we have a whole service program we've, talked, about it previously on, the podcast but I think it's changed a little since then which is kind of cool so if you're a start-up and you're like oh man I really want to use a Google cloud for it you can go to cloud.google.com such. Startups and. It'll show you the startup, page which is here and there are different tiers that are available right now for different, types of startup. Packages so you can get anywhere from three, thousand dollars in credit all the way up to hundred thousand dollars in credit for one year including. As well like 24/7. Support office, hours. Spotlights, like all sorts of different things depending on what tier you're on and all sort of needs you have there, are several actually we've had discard, eckhart labs which have been on the podcast more than once is one of those people who came through that program the, one thing to know though if you are startup and you're looking and, you want to come through this program you need to apply through your incubator accelerator or, VC firm there's. A long list, of all those people that we work with our platform partners, I counted. 142. Nice, yeah, it's a good number of people to be working with so, yeah if you're a small business and you're looking for some credits and it gives some insights, in terms of like the size what that means a little bit of information around who would qualify for the program. Check, out our link, you'll see that there's start, package there's a spark package, and there's a surge package, in terms of the number of credits that you can potentially get hold of and this. Can be applied to all of GC and firebase yes, actually that's a really good point that it goes to both TCP and firebase yes which is also super cool if, you aren't part, of one of those groups there's incubators or accelerated programs there is a mailing list you can sign up for just, to get updates to see if anything changes or if new people come on board that's, there and obviously free, trial is also obviously, available there, we go well, that wraps, us up for at least our first interview out the gate for 2018 anything. You want to actually share, or any places you're gonna be in the next couple weeks I'm not, traveling very much queue ones for me is very quiet normally, I'm just sitting around trying to get things done for Game Developers Conference and March just trying to get ramped up for that how about yourself you off anywhere nothing to announce right now but probably, later yeah all, right cool well Melanie thank, you for joining me for the, first episode of the year of 2018.

Done, Yeah, it's done and thank you all for listening and we'll see you all next week.

2018-01-13 16:14

Show Video

Other news