>> Announcer: Live from Stanford University, it's theCUBE, covering the Women in Data Science Conference 2017. >> Hi, welcome to theCUBE. I'm Lisa Martin, and we are at the Women in Data Science second annual conference at Stanford University. Great event, very excited to be joined by one of the founders of the Women in Data Science, the Senior Director and Head of Data Science at Walmart Labs, Esteban Arcaute. Very nice to have you on the program. Thanks for joining us. >> Thank you for having me, Lisa. >> So talk to us about data science in retail. How is Walmart using data science too influence shoppers wherever they are, mobile, in store, dot com? >> So data science is a key component to how we create our experiences, especially now that our customers essentially don't really make a distinction between they're shopping in stores or they're actually using their mobile device, or they're at home with their desktop. So that means that for us it really is about creating a seamless experience that allows a customer to not feel that barrier of the medium that they're using to shop. So more practically, that means that the data that we're using to create the experience is essentially the same across all of these medias. >> So big data brings, and data science brings big opportunities, but also some challenges. Talk to us about some of the challenges that you've had with the tremendous amount of data because you've got what? Sixty million shoppers, 260 million, excuse me, globally. How are you dealing with some of those challenges and really turning them into opportunities to create that seamless experience? >> So for us it means that a lot of ready-made solutions that are available for other companies, they just don't work for us. The same way that other companies with large amounts of data, they actually have to create their own in-house solutions or technology. It is the same for us. Now in terms of how that is a very specific challenge, that means that when you actually go and train, let's say a model, that is trying to predict whether a customer is going to satisfied with a purchase or not, usually the amount of data that you have will make that model to not be that reliable unless you actually did it in-house. >> Okay, so from an accuracy perspective that really is what was driving being able to do that within Walmart Labs? >> Yes, and just sort of to give a plug to the department where I got my PhD, all of these numerical instabilities that in past you will only see when doing computational fluid dynamics, they actually start appearing in places like retail just because of the volume of data that is available. And so for us it's a great opportunity to be an ICME student. >> Excellent, and that's right, you got your Master's and your PhD right here at Stanford. Talk to us about from a scale and a speed perspective. How are you seeing the ability to influence the consumer experience? How quickly are you able to identify trends and act on them so that customer experience is better, and also the bottom line financials are improved as well for Walmart? >> That is a great question, Lisa, because our customers' expectations are changing really, really rapidly. If you remember back in the late 90s when you would go to a search engine and it worked, it was like a miracle. Everybody was really excited. Fast forward to today, you go to any search box, not a search engine, you put in a query. If it doesn't work, you're disappointed. When it works, it's just table stakes. That means that for us we need to be able to iterate as quickly as the customer expectations change, which is really, really fast. >> Absolutely. How do you collaborate with the business side? So first, let's talk about your team. What's the size of your team? As the head of data science, what are the different functions within your team, first and foremost? >> I'm also in charge of the search experience within Walmart Global eCommerce. It's a fairly large team because it is composed of basically the full stack from the back end, data science, dev ops, product management, so I cannot give you an exact size, but it's a fairly large team. >> And so how do you collaborate with the business to influence merchandising, for example? What is that collaboration like between Walmart Labs and the dot com side? >> So last year, Kelly Thompson was one of the speakers at the Women in Data Science Conference, and she talked about the importance of bringing the art of merchandising with the science of data science together. And it really is true that there're certain things that algorithms cannot catch as soon as a human expert actually knows about. And so the way we develop our products and enhance experiences for our customers is really bringing these two together in a partnership to ensure that there's never one side that is working on something that the other one cannot just leverage. >> From a priority perspective, how are some of the trends that you find driving priorities for investment? >> It goes both ways. Sometimes we find the trend. Sometimes the business finds the trend. And so sometimes the business asks us to try to automate or to predict something that we hadn't thought about, and that is actually very difficult, and hence we invest a lot in that. And sometimes we find some customer patterns that indicate a different behavior in a locality or with certain characteristics that then the business can go and better serve themselves. So it really is driven by whoever has a good idea, and they can come from anywhere. >> You mentioned the need still for human insight. Talk to us about that dynamic, machine learning and human insight. How does that work together, and again kind of thinking in the context of speed and skill to meet those changing customer demands? >> That is one of the best kept secrets for machine learning, is that most machine learning systems, the moment they have a human in the loop, the learning grade gets accelerated exponentially bcause essentially when a machine learning method is not working properly, it tends to be for certain types of cases that if they get resolved, just a few insights from a human being can actually go and make the machine learn a lot faster than if it's trying to figure it out on its own. So for us really even there is a partnership. We think of it as a systems with a human in the loop. That human, if it's an expert, it's even better, which is what we have. And so we create our systems to deeply integrate our merchandising capacity. >> So you actually see human intervention or interaction as a necessary component to speed to market leveraging data? >> That is the fastest way to get there. There might be other ways to do with that. We don't always have a human in the loop, but when we can have a human in the loop, we have seen that acceleration is actually measurable. >> Fantastic. So one of the things I wanted to chat about with you is looking at your team a little bit, as well as your involvement here in the Women in Data Science. You were one of the founders. Talk to us about Walmart's interest in helping to not only educate women, and further their education in data science, but also maybe to combat the predicted shortage of data scientists that's predicted to start even in 2018. How is that collaboration going to help in that sense? >> So let me address the question in two parts. First, the question related to women and minorities into data science. So Walmart is a very inclusive company. We win awards every year because of all of our work in there. And I think that starting with Women in Data Science, it's a natural place to start because there's always 50% of women everywhere. And so that means that really thinking that there should be an equal representation, or maybe not equal representation, there should be a way to funnel all of this talent into data science just makes sense. There's not a question as to whether there's sufficiently many of them or things like that. >> So culturally it was kind of a natural extension for Walmart Labs it sounds like. >> Absolutely, yes. And the second question is the shortage. So for us we're very lucky in that we have two things that any company needs to have to attract great data scientists. So first one is that we actually have data. Believe it or not, it is an asset that a lot of companies don't realize is actually (mumbling). And the second one is that we empower all of our associates with the ability to have impact from the get go. We don't put them in some small project that might have an impact in maybe three years. No, we actually put them in participating projects that might have, for instance in my team, impact within the first three to four months of being on the floor. >> That's fantastic, and I'm sure that really inspires them. They see that they can make an impact right away. And I would imagine just after chatting with you that they have the freedom probably to test and fail, and from that failure it becomes more opportunities to get and tweak and get things right. >> Absolutely. So especially in a field like retail, there's no laws of retail. There's not someone that just put in some nice equations and we just and study and do something. Actually you need to test over and interate constantly, especially when your customers expectations change so rapidly. >> So in terms of evolution of data science and skills, data presentation skills, analysis, stats, math, what are some of the other skills, maybe even social skills that you think are really key for the young next generation of data scientists to really get into this field regardless of industry and be successful? >> It's a question that I get very often, and especially because data science has not yet been formally properly defined in some sense. Data scientist is even less properly defined, so the term just started in 2010 or 11, so usually people think that they have to be hackers, have analytical skills and have some domain expertise. We actually flip that to say you have to have analytical skills, so that stays. You have to be a software engineer or have software engineering skills, and you have to project management skills. And the reason is that unless you are able to properly communicate what your insights are, to understand how they get incorporated into a real software system, and of course to have the expertise to know what you are doing, you're not going to be successful as a data scientist. So for us really those three components are the ones that drive what are we looking at data scientists. >> Excellent, so you mentioned hackers. Hackathons, you recently had a hackathon. How is Walmart Labs giving opportunities to maybe kids in grade school and high school, kids that are university to start developing that talent. >> So we have also an internship program every year. We have interns across all of Walmart Labs, and there is always a great opportunity to seed fresh new ideas that come from our interns, so that happens every year. We organize hackathons in very targeted way in places where we see that there is demand to have these kind of events organized. So I think one that we have in our website is one from 2015 with Tech Crunch Disrupt. It's a big one, but we do other things as well. >> But that actually has the ability, someone who's made a big difference or won at a hackathon that Walmart Lab sponsors has the ability to actually influence Walmart. >> Absolutely because as I said a couple of minutes ago, great ideas come from anywhere. And hackathons are great places where you see all of these ideas bubbling, and that you might not even realize that oh, that opportunity is right there. Someone can see it, and wants it seen, everybody can see it. So it's a great place. >> But that's a great, from a cultural perspective what you're saying sounds fantastic, that you're, there's a culture within Walmart Labs and Walmart that really is not only diverse from women in the sciences as well, but also one that really encourages test it, try it, you can make an impact here. And I think that's huge for attracting talent. What advice would you give to some of the young women that are here at the Women in Data Science Conference for the second annual to want to become successful data scientists? >> So I would give the advice that I have for myself, which is stay true to yourself, and anyone can be a great data scientist. >> What are some of the things that you're most looking forward to learning and hearing at this second annual event? >> The line up of speakers is amazing, and I think that the fact that they come from all places in industry, and all types of academic and professional journeys make it a very rich experience even for me to understand what are the possibilities. >> Absolutely, the cross section of speakers at the event is amazing. You've got obviously you know, data science into retail. We've got people that are using, that are going to be on the show later, data science to change the way college kids are recruited for jobs. Kind of getting away from that things that used to scare me, GPA, test scores, really leveraging science to open up those possibilities. And I think one of the things that that can enable from your comment earlier is the importance of being able to be a good communicator. It's not just about understanding the data. You've got to be able to explain it in a way that makes sense. Is this an impact? Also you mentioned we've got people that are here today on the academic side that are helping to educate the next generation of computer and data scientists. So I think it's a phenomenal opportunity for women of all ages to really understand it's not just technology. Every company this day and age is a technology company, and the opportunities are there to be influencers, and it sounds like at Walmart Labs, from the ground up. >> Yes, absolutely. >> Fantastic. Well, Esteban it's been such a pleasure having you on the program today. Thank you so much for joining. We look forward to having a great event and hopefully seeing you at the third annual next year. >> Definitely. Thank you very much for having me, Lisa. >> And you've been watching theCUBE. We are live at the Women in Data Science Conference at Stanford University. Stick around, be right back. (jazzy music)