 Live from Orlando, Florida, it's theCUBE. Covering Microsoft Ignite, brought to you by Cohesity. Good afternoon, everyone, and welcome back to theCUBE's live coverage of Microsoft Ignite, 26,000 people here at this conference at the Orange County Convention Center. I'm your host, Rebecca Knight, alongside my co-host, Stu Miniman. We are joined by Seth Juarez. He is the Cloud Developer Advocate at Microsoft. Thank you so much for coming on the show. I'm so glad to be here. You have such a lovely set and you're lovely people. Thank you very much, Seth. All right. You just met up, you don't know any better. No, well, maybe after the end of the 15 minutes, we'll have another discussion. You're starting off on the right foot. I love it. So, tell us a little bit about what you do. You're also a host on Channel 9. I am. Tell us about your role as a Cloud Developer Advocate. So, a Cloud Advocate's job is primarily to help developers be successful on Azure. My particular expertise lies in AI and machine learning, and so my job is to help developers be successful with AI in the cloud, whether it be developers, data scientists, machine learning engineers, or whatever it is that people call it nowadays, because you know how the titles change a lot, but my job is to help them be successful. And sometimes what's interesting is that sometimes our customers can't find success in the cloud. That's actually a win for me too, because then I have a deep integration with the product group, and my job is to help them understand from a customer perspective what it is they need and why. So, I'm like the ombudsman, so to speak, because the product groups are the product groups, I don't report up to them, so I usually go in there and I'm like, hey, I don't report to any of you, but this is what the customers are saying. We are very keen on being customer centered, and that's why I do what I do. Seth, I have to imagine when you're dealing with customers, some of that skills gap in learning is something that they need to deal with. You know, we've been hearing for a long time, you know, there's not enough data scientists. You know, we need to learn these environments. Satya Nadella spent a lot of time talking about the citizen developers out there. So, you know, bring us inside the customers you're talking to, you know, kind of where to usually start, and you know, how do they pull the right people in there, or are they bringing in outside people? A little bit of organization. It's a great question. It turns out that for us, at Microsoft, we have our product groups, and then right outside we have our advocates that are very closely aligned to the product groups, and so anytime we do have an interaction with a customer, it's for the benefit of all the other customers. And so, I meet with a lot of customers, and I don't get to talk about them too much, but the thing is, I go in there, I see what they're doing. For example, one time I went to the Turing Institute in the UK, I went in there, and because I'm not there to sell, I'm there to figure out, like, what are you trying to do, and does this actually match up? It's a very different kind of conversation, and they tell me about what they're working on, I tell them about how we can help them, and then they tell me where the gaps are, or where they're very excited, and I take both of those pieces of feedback to the product group, and they just love being able to have someone on the ground to talk to people, because sometimes, you know, when you work on stuff, you get a little siloed, and it's good to have an ombudsman, so to speak, to make sure that we're doing the right thing for our customers. As somebody that works on AI, you must've been geeking out working, working with the Turing Institute, though. Oh yeah, those people are absolutely wonderful, and it was like, as I was walking in, a little giddy, but the problems that they're facing in AI are very similar to the problems that people, other people doing AI that are in big organizations. Other organizations are trying to onboard to AI and try to figure out, everyone says they may be using this hammer, and they're trying to hammer some screws in with the hammer, so it's good to figure out when it's appropriate to use AI and when it isn't, and I also have customers with that. And I'm sure the answer is it depends in terms of when it's appropriate, but do you have any sort of broad brush advice for helping an organization determine, is this a job for AI? Absolutely, it's a question I get often, and developers, we have this thing called a smell that tells us if a code smell, we have a code smell, it tells us maybe we should refactor, maybe we should, for me, there's this AI smell where if you can't precisely figure out the series of steps to execute an algorithm and you're having a hard time writing code, or for example, if every week you need to change your if-else statements, or if you're changing numbers from point five to point seven and now it works, that's the smell that you should think about using AI or machine learning, right? There's also a set of, a class of algorithms that, for example, AI, it's not that we've solved them, but they're pretty much solved, like for example, detecting what's in an image, understanding sentiment and text, right? Those kinds of problems we have solutions for that are just done, but if you have a code smell where you have a lot of data, and you don't want to write an algorithm to solve that problem, machine learning and AI might be the solution. All right, a lot of announcements this week, any of the highlights from your area? We, last year, AI was mentioned specifically many times, now it's autonomous systems, and it feels like AI is in there, not necessarily just rubbing AI on everything. I think it's because we have such a good solution for people building custom machine learning that now it's time to talk about the things you can do with it. So when we're talking about autonomous systems, it's because it's based upon the foundation of the AI that we've already built. We released something called Azure Machine Learning, a set of tools in a studio where you can do end-to-end machine learning because what's happening is most data scientists nowadays, and I'm guilty of this myself, we put stuff in things called Jupyter Notebooks, we release models, we email them to each other, we're emailing Python files, and that's kind of like how programming was in 1995. And now what we're doing is we're building a set of tools to allow machine learning developers to go end-to-end, be able to see how data scientists are working, et cetera. For example, let's just say your data scientist, Bill, did an awesome job, but then he goes somewhere else, and Sally, who is absolutely amazing, comes in, and now she's the data scientist, usually Sally starts from zero, and all of the stuff that Bill did is lost. With Azure Machine Learning, you're able to see all of your experiments, see what Bill tried, see what he learned, and Sally can pick right up and go on, and that's just doing the experiments. Now, if you want to get machine learning models into production, we also have the ability to take these models, version them, put them into a CICD similar process with Azure DevOps and machine learning, so you can go from data all the way to machine learning and production very easily, very quickly, and in a team environment, and that's what I'm excited about mostly. So at a time when AI and technology companies in general are under fire, and not considered to not always have their user's best interest at heart, I'd like you to talk about the Microsoft approach to ethical AI and responsible AI. Yeah, I was a part of the keynote, Scott Hansen was a very famous dev, and he did a keynote, and I got to form part of it, and one of the things that we're very careful, even on a dumb demo where he was like doing rock, paper, scissors, I said, and Scott, we were watching you with your permission to see what sequence of throws you were doing. We believe that through and through all the way. We will never use our customer's data to enhance any of our models. In fact, there was a time when we were doing like a machine learning model for NLP, and I saw the email thread, and it's like, we don't have language foo, I don't remember what it was. We don't have enough language foo. Let's pay some people to ethically source this particular language data. We will never use any of our customer's data, and I've had this question asked a lot, like for example, our cognitive services which have built in AI, we will never use any of our customer's data to build that. Neither, for example, if we have, for example, we have a custom vision where you upload your own pictures, those are your pictures, we're never going to use them for anything, and anything that we do, there's always consent, and we want to make sure that everyone understands that AI is a powerful tool, but it also needs to be used ethically, and that's just on how we use data for people that are our customers. We also have tools inside of Azure Machine Learning to get them to use AI ethically. We have tools to explain models, so for example, if you vary gender, does the model change its prediction, or if you vary class or race, is your model being a little iffy? We allow, we have those tools in Azure Machine Learning so our customers can also be ethical with the AI they build on our platform. So we have ethics built into how we build our models, and we have ethics built into how our customers can build their models too, which is to me very exciting. And is that a selling point? Our customers gravitating, I mean, we've talked a lot about it on this show about the trust that customers have in Microsoft, and the image that Microsoft has in the industry right now, but the idea that it is also trying to perpetuate this idea of making everyone else more ethical, do you think that that is one of the reasons customers are gravitating? I hope so, and as far as a selling point, I absolutely think it's a selling point, but we've just released it, and so I'm going to go out there and evangelize the fact that not only are we ethical with what we do in AI, but we want our customers to be ethical as well, because you know, trust pays. As Satya said in his keynote, trust is the enhancer in the exponent that allows tech intensity to actually be tech intensity, and we believe that through and through, not only do we believe it for ourselves, but we want our customers to also believe it and see the benefits of having trust with our customers. Yeah, one of the things we talked to Scott Hanselman a little bit yesterday about that demo is the Microsoft of today isn't just use all the Microsoft products, it's allowed you to use any tool, any platform, your own environments. Tell us how that plays into your world. You know, in my opinion, and I don't know if it's the official opinion, but we are in the business of renting computer cycles. We don't care how you use them. Just come into our house and use them. You want to use Java. We've recently announced tons of things with Spring. We've become an open JDK contributor. You know, one of my colleagues worked very hard on that. I work primarily in Python because it's machine learning. I have a friend and colleague, David Smith, who works in R. I have other colleagues that work in a number of different languages. We don't care. What we are doing is we're trying to empower every organization and every person on the planet to achieve more where they are, how they are, and hopefully bring a little bit of it to our cloud. What are you doing that that's really exciting to you right now? I know you're doing a new .NET library. Any other projects that are sparking your interest? Yeah, so next week I'm going to France, and this is before anyone's going to see this, and there is a company, I think it's called Surf, I'll have to look it up and we'll put it in the notes, but they are basically trying to use AI to be more environmentally conscious, and they're taking pictures of trash and rivers, and they're using AI to figure out where it's coming from. So they can clean up the environment. I get to go over there and see what they're doing, see how I can help them improve it, and promote this kind of ethical way of doing AI. We also do stuff with snow leopards. I was watching some Netflix thing with my kids, and we were watching snow leopards, and there was like two of them, like this is impressive, because as I'm watching this with my kids, I'm like, hey, we are at Microsoft, we're helping this population perpetuate with AI, and so those are the things I'm excited about. It's actually, and how I've seen on TV is, rather than spending thousands of hours of people out there, the AI can identify the shape through the camera, so there, I love that, it's a powerful story to explain some of those pieces as opposed to, it's tough to get the nuance of what's happening here. Absolutely. With this technology. And those models are incredibly easy to build on our platform, and I say incredibly easy to build with what you have. We love, people use TensorFlow, use TensorFlow. People use PyTorch, that's great. Cafe, Teana, whatever you want to use, we are happy to let you use or rent out our computer cycles because we want you to be successful. Yeah, maybe speak a little bit of that when you talk about the cloud, one of the things is to democratize availability of this. There's usually free tiers out there, especially in the emerging areas. How's Microsoft helping to get that compute and that first world technology to people that might not have had it in the past? I was in Peru a number of years ago, and I had a discussion with someone on the Channel 9 show, and it was absolutely, I suddenly understood the value of this. He said, Seth, if I wanted to do a startup here in Peru, right, and it was a capital Peru, like a very industrialized city, I would have to buy a server. It would come from California on a boat. It would take a couple of months to get here, and then it would be in a warehouse for another month as it goes through customs, and then I would have to put it into a building that has AC, and then I could start. Now, Seth, with a click of a button, I can provision an entire cluster of machines on Azure and start right now. That's what the cloud is doing in places like Peru, in places that maybe don't have a lot of infrastructure. Now, the infrastructure is for everyone, and maybe someone even in the United States, you know, in a rural area that doesn't, they can start up their own business right now, anywhere. And it's not just because it's Peru, it's not just because it's some other place that's becoming industrialized. It's everywhere because any kid with a dream can spin up an app service and have a website done in like five minutes. So what does this mean? I mean, as you said, any kid, any person in a rural area, any developing country, what does this mean in five or 10 years from now in terms of the future of commerce and work and business? Honestly, some people feel like computers are stealing human engineering. I think they're really augmenting it. Like, for example, I don't have to, if I want to know something, remember back when I was a kid, I had to, if I wanted to know something, sometimes I had to go without knowing. We're like, I guess we'll never know, right? And then five years later, we're like, okay, we found out. It was that a character on that show, you know? And now we just look at our phone and it's like, oh, you were wrong. And I like not knowing that I'm wrong for a lot longer. You know what I'm saying? But nowadays with our phones and with other devices, we have information readily available so that we can make appropriate response, appropriate answers to questions that we have. AI is going to help us with that by augmenting human ingenuity by looking at the underlying structure that we can't. For example, if you look at Excel spreadsheet, if it's like five rows and maybe five columns, you and I as humans can look at it and see a trend. But what if it's 10 million rows and 5,000 columns? Our ingenuity has been stretched too far, but with computers now, we can aggregate, we can do some machine learning models, and then we can see the patterns that the computer found aggregated and now we can make the decisions we can make with five columns, five rows. But it's not taking our jobs, it's augmenting our capacity to do the right thing. Excellent, well Seth, thank you so much for coming on theCUBE, a really fun conversation. I'm so glad to be here, thanks for having me. All right, I'm Rebecca Knight for Stu Miniman. Stay tuned for more of theCUBE's live coverage of Microsoft Ignite.