>> live from London, England. It's the queue covering a ws summat London twenty nineteen, Brought to you by Amazon Web services. >> We're at the A. W s summits here in London, at the XL Center there are thousands and thousands of delegates here looking to see the future for their own technologies on what Kyle will hold for them, as well as lots of the other established players here. There are plenty of startups. I'm just down the street and this is my co host, Dame Ellen. We're gonna be talking to a few of the startup founders who are with us here on the Cuban. It's great to have you here. So first up, Hu Pei jin, Who is that? The co founder of the three d mapping based service. And this is called Escape Technologies, but also chanting Jane. And you are the co founder. A swell founder, I believe is it found it found in co founder ofyour organization called Kwan. Now let me festival starts talking to Jan Di and about what you do because you're offering a service to financial services. Are you on helping them with machine learning? Teo, try and offer the best portfolio managers for wealth investment. How does it work? What you're offering? >> Yes, our platform basically allows traders, portfolio managers, asset managers who want to make smarter investment decisions to build machine learning models. To do this Theo idea is that data driven investing should help funds make more profits for themselves and their clients. But there's not enough data, scientists, King data scientist who can actually do more good for them. And we address this lack of talent by using a community of data scientist people who come from outside of finance to help them crowd to help fund managers crowdsource model, using their intelligence, their talent. So the process is really simple. Clients come to us with what we like to call an investment problem or a finance problem. We take that problem and convert it into a pure matter. And she learning problem. That's someone who is not from finance, can understand and soil >> so really interesting. You say that because I've spoken to other founders of other data companies who say, for example, be looking at the stars for their main bread and butter. But then Khun transfer those skills and astronomy to the financial sector and those types of people that you're trying to harness their skills. >> Yeah, exactly. So our community is made up of people who work at tech. Companies at Google and Amazon have sport off people who are putting graduate program and computer science and math machine learning, but don't necessarily know finance. And the idea is, can you make this problem than two problems? Can you make finance problem into problems that this community of data scientists really smart data scientists understand without needing to know finance? >> It's interesting that it lord, because ofthe a lack of of data scientists, Really? But do you think if you eliminate all the kind of heavy lifting out of what you do in the future, though, will be a need for fewer data? Scientists? >> I don't think we need to fut the scientist, but they wouldn't be a need for reform Toe have in house teams. They will basically be able to. A data scientist working in an unequal miss company should be able to solve problems of a finance company. The scientists working in uber should be able to solve problems for a hedge fund because we're building this translator that can allow knowledge from anywhere to be used to solve any kind of problems. >> Okay, let me talk to you because you do three d mapping services. Why do you think these are essential for technologies large and small? Going forward, >> Esso and every future industry in the future is going to have some autonomous aspect to it. So if you think about Atanas vehicles, ever think about delivery Jones. These are going to be machines. They're going to be acting autonomously in human like environments, and they're going to make decisions based on purely what they're observing with hardly human in between. So the only way that this can happen intelligently and safely is if those machines also have a human like understanding ofthe human like environment, just like you humans. So while we are providing these things, machines with Is that human like understanding and the first service that we're building towards that is a visual positioning system to provide the machines with the ability to answer the question. Where am I now? The only way that you can provide official positioning system is this. If you also have a visual map off of the world on this math needs to be updated in real time. So for every future industry, having a real time update version off the real world is fundamental. That's the pinnacle around. Every single every single decision that autonomous agent is going to make is going to be based upon this map. >> So this map was really value Peace Corps piece, um, that we're building. So I've often wondered if people talk about autonomous cars, but we don't have things like autonomous cart's right now. People will say, Well, an Amazon warehouse would have that. But there, following beacons or stripes, Yeah, what you're talking about is potentially taking >> us to the point where you can break that barrier. Is that fair? Exactly. And for warehouses, I would forever advice to use those beacons. Because warehouses are pre pre massaged environments, you define what the environment looks like. Whereas humans we walk around in cities, in nature and all these places that are not pre processed, we have to take our cues from the visuals that we observe. So if you go back to your hometown, for example, you observe a Starbucks logo Starbucks logo and observe our street sign, you might be able to very opposition based on those visual visual cues. Even though the environment itself was not pre processed to provide those cues, the cues are already in the nature. So >> we've heard that there have bean in these trials that have bean accident. There's a limit that is >> Oh, yeah, totally. So at the moment, they're sure are accidents, But you are a human. You can navigate properly with any human environment, using your visual sense it your eyes. Therefore, any machine will, in the future only need that visual sensor as well. So only a camera to navigate around the world were seeing great great progress on the neural networks, deep learning as well as on the geometry and visual image processing, like the type of computer vision that we do that are making so much progress that guaranteed a couple of years from now, the devices will have the understanding off the world like humans do. And we'LL be able to make decisions even better than humans do because they don't got there. They don't get tired. They don't need coffee. S o. B. Guaranteed. More safe than any human knowledge. It's Sunday, and you probably hate the term robo investing, right? But but it sounds like you're doing that form of machine investing for and with hedge funds is that isn't fair. And is your background finance data science or both? >> Both. Actually, I studied engineering, but I started working as a trader of infidelities trading company in Chicago. On that I started with them. We were very old school discretionary, you know, a couple of very senior guys who were making everything based on their past experience and that contusion about the market. On my time with them, he started shifting from this manual human process driven trading to something that was more systematic, inconsistent again. That's where the whole idea >> for all >> Kwan came from. I saw firsthand the benefits that making your trading more data driven more model and algorithms driven could have >> unique. You probably hate this trump to your unicorn, but I'm guessing you guys have no it shop is You're right. It is in the cloud. Is that writer OK, >> it is, you know, straight onto the cloud todo in that started. You didn't exist before. >> Yeah, yeah, Waylon Street in the club. >> And you got a team of developers. They program infrastructure. Totally. >> Yeah. We have a team off for developers and the city of totally tech team of five based out of India. We have a developed sky who basically runs everything for us. Our website, Our platform where the data scientist party prision where our clients see the mortals where client fronts for data to us and where our machine learning computations run >> right three t mapping used to buy a box the Unix box, maybe get a database mother software. Yeah, so we're in scale were thought of as well, right? So when we what you need is the process. If you want to create a three d map off even a city but we have to do is run eight hundred GPS in parallel, blasting through imagery data. Now, this is impossible. If we as a starter had to buy a GPU wreck right from the bat, we would have been bankrupt even before we started. So, like being able to spin up GPU servers in the cloud and also killing them after we're done with them say there's a lot of money but also provides so much flexibility for us to do prototyping and two on DH to make everything affordable and east implement with very, very small team of very talented system. >> It's a real kind of pick and mix approach. Just what kind of services do I need to get off the shelf? And then it happened to you? >> I think one of the great things that a US has been able to do infrastructure used to be a very dusty and tangled industry on one of the beauties that Davy was able to do is actually product eyes, product, eyes, infrastructure. So you can now actually pick and choose different products from the idea of a library and put them together, connect them, tied him up very, very cleanly. With a very small team, I create something that is just accedes. Any expectations from a start of twenty years ago. So why, why eight of us? A lot of other clouds out there who has got a good cloud. Microsoft has a big cloud. Why did you guys migrate or moved to eight of us not moved to start with a W s. How was that decision made? >> I mean, we started with eight of us because we were gonna start a program a date afterwards. But then we just really liked the support that we got a way. We had access to someone twenty four seven. We had a dedicated person who was helping us on DH. We were just starting out. So the first time interacting with a cloud infrastructure, uh, the support was greater than the pricing will go great. For a start, it would have to say that's just a start of ur cost sensitive and the ability to turn on on and off services as and when we need them. I think that was fantastic. >> Does it concern you that we've heard a lot about how the cost of services has come down quite a lot? There's a lot of Costco going, but in the future, if you're overly reliant on your provider, can that put you into a corner? >> I mean, you get into troubles if your spotify skill, but as a start of the environment that ate us created for startups to flourish, is incredible. The amount of I think you have the same, like we receive a huge amount of credits just for starting. So if you raise a seed round of money which is, let's say, one million U. S. Dollars. US puts one hundred thousand worth of credit. On top of that, that's ten percent extra funding for free provided. Wait. Oh, yes. Furthermore, they have this great architects. The help you out with all the questions that you might have if this is the first time that you are actually designing a whole our detector around a data processing apartment or an FBI or a Web platform? Very, very supportive. What was that? What's the one thing a ws would could do to make your life easier? If you're sitting here with Andy Jassy, what would you tell him? >> I mean, it's already fantastic. It's made our life so much easier. I really don't think of anything that could have gone better. >> Really? Nothing. I mean, you had reduced the cost even way prices. >> Okay. Well, thank you so much for talking to us about your experiences here on the Cube. Who? Heiner. Thank you. Co founder of Escape. And also it'LL really, Jane, it's really be fascinating to hear how you've grown your businesses. So I really appreciate you joining us here with me. Damayanti here at eight Ws summits in London