 Live from Washington D.C., if the Cube. Covering Inforum DC 2018, brought to you by Infor. Well, welcome back to Washington D.C. We are live here in the convention center at Inforum 18, along with Dave Vellante. I'm John Walls. It's a pleasure now. Welcome to the Cube. Lee Martin, who is the senior director of the Dynamic Science Labs at Infor. And good afternoon to you, Lee. Good afternoon. Thank you for having me. Thank you for being here. All right, well, tell us about the labs first off. Obviously, data science is a big push at Infor. What do you do there? And in why is data science such a big deal? Yeah, so Dynamic Science Labs is based in Cambridge, Massachusetts. We have about 20 scientists with backgrounds in math and science areas. So typically PhDs in statistics and operations research in those types of areas. And we've really been working over the last several years to build solutions for Infor customers that are math and science based. So we work directly with customers, typically through a proof of concept. So we'll work directly with customers, we'll bring in their data, and we will build a solution around it. We like to see them implement it and make sure we understand that they're getting the value back that we expect them to have. And then once we prove out that piece of it, then we look for ways to deliver it to the larger group of Infor customers, typically through one of the cloud suites, perhaps functionality that's built into a cloud suite or something like that. Well, give me an example. I mean, so as you think, you're saying that you're using data that's math and science based for application development or solution development, if you will. Like how? So I'll give you an example. So we have a solution called Inventory Intelligence for Healthcare, which we're, it's moving towards a more generalized name of inventory intelligence because we're going to move it out of the healthcare space and into other industries. But this is a product that we built over the last couple of years. We worked with a couple of customers. We brought in their loss in data, so their loss in customers. We bring the data into an area where we can work on it. We have a scientist in our team actually, she's one of the senior directors in the team, Don Rose, who led the effort to design and build this design and build the algorithm underlying the product. And what it essentially does is it allows hospitals to find the right level of inventory. Most hospitals are overstocked, so this gives them an opportunity to bring down their inventory levels to a manageable place without increasing stockouts. So obviously it's very important in healthcare that you're not having a lot of stockouts. And so we spent a lot of time working with these customers, really understanding what the data was like that they were giving to us. And then Don and her team built the algorithm that essentially says, here's what you've done historically. So it's based on historic data. At the item level, at the location level, what have you done historically and how can we project out the levels you should have going forward so that they're at the right level where you're saving money, but again, not implementing, you're not increasing stockouts. So it's a lot of time and effort to bring those pieces together and build that algorithm and then test it out with the customers. You try it out a couple of times. You make some tweaks based on their business process and exactly how it works. And then, like I said, we've now built that out into originally a standalone application and in about a month, we're going to go live in Cloud Suite Financial, so it's going to be a piece of functionality inside of Cloud Suite Financial. So, John, if I may, I'm going to digress for a moment here because the first data scientist that I ever interviewed was the famous Hillary Mason, who's, of course, now at Cloudera, but and she told me at the time that the data scientist is part mathematician, part scientist, part statistician, part data hacker, part developer and part artist. And so, you know, it's an amazing field that Hal Varian, who was the Google economist said, you know, it's going to be the hottest field in the next 10 years. And it's sort of proven true, but to leave my question is, so you guys are practitioners of data science and then you bring that into your product. And what we hear from a lot of data scientists, other than that sort of, you know, Penelope of skill sets, is they spend more time wrangling data and the tooling isn't there for collaboration and how are you guys dealing with that? How has that changed inside of Infor? It is true. And we actually really focus on first making sure we understand the data and the context of the data. So it's really important if you want to solve a particular business problem that a customer has, to make sure you understand exactly what is the definition of each and every piece of data that's in all of those fields that they send over to you before you try to put them inside an algorithm and make them do something for you. So it is very true that we spend a lot of time cleaning and understanding data before we ever dive into the problem solving aspect of it. But, and to your point, there are, there is a whole list of other things that we do after we get through that phase. But it's still something we spend a lot of time on today and that has been the case for, you know, a long time now. We, wherever we can, we apply, you know, new tools and new techniques, but actually just a simple act of going in there and saying, what am I looking at? How does it relate? Let me ask the customer to clarify this and make sure I understand exactly what it means. That part doesn't go away because we're really focused on solving the customer solution and then making sure that we can apply that to other customers so really knowing what that data is that we're working with is key. So I don't think that part has actually changed too much. There are certainly tools that you can look at and people talk a lot about visualization so you can start thinking, okay, how can I use some visualization to help me understand the data better? But just that whole act of understanding data is key and core to what we do because we want to build the solution that really answers the business problem. The other thing that we hear a lot from data scientists is that they help you figure out what questions you actually have to ask. So it sort of starts with the data, you analyze the data, maybe you visualize the data as you just pointed out and all these questions pop out. So what is the process that you guys use? You have the data, you've got the data scientist, you're looking at the data, probably asking all these questions, of course I'm getting questions from your customers as well. You're building models maybe to address those questions, training the models to get better and better and better and then you infuse that into your software. So maybe is that the process? Is it a little more complicated than that? Maybe you could fill in the gaps. So my personal opinion and I think many of my colleagues would agree with me on this starting with the business problem for us is really the key. There are ways to go about looking at the data and then pulling out the questions from the data but generally that is a long and involved process because it takes a lot of time to really get that deep into the data. So when we work we really start with what's the business problem that the customer's trying to solve and then what's the data that needs to be available for us to be able to solve that and then build the algorithm around that. So for us it's really starting with the business problem. Okay, so what are some of the big problems? We heard this morning that there's a problem in that there's more job openings than there are candidates and productivity. Business productivity is not being impacted. So there are two big chewy problems that data scientists can maybe attack and you guys seem to be passionate about those. So how does data science help solve those problems? So I think that at INFOR, I'll start off by saying at INFOR there's actually, I talked about the folks that are in our office in Cambridge but there's quite a bit of data science going on outside of our team and we are the data science team but there are lots of places inside of INFOR where this is happening. Either in products that contain some sort of algorithmic approach, the HCM team for sure, the talent science team which works on HCM. That's a team that's led by Jill Strange and we work with them on certain projects in certain areas. They are very focused on solving some of those people related problems. For us we work a little bit more on the, some of the other areas we work on is sort of the manufacturing and distribution areas. We work with the healthcare side of things. So supply chain, healthcare. Exactly, so some of the other areas. Because there are, like I said, there are some strong teams out there that do data science, it's also incorporated with other things like the talent science team. So there's lots of examples of it out there. In terms of how we go about building it, so we, you know, like I was saying, we work on answering the business question up front, understanding the data, and then really sitting with the customer and building that out. And so the problems that come to us are often through customers who have particular things that they want to answer. So a lot of it is driven by customer questions and particular problems that they're facing. Some of it is driven by us. We have some ideas about things that we think would be really useful to customers. Either way, it ends up being a customer collaboration with us, with the product team, that eventually we want to roll it out to, to make sure that we're answering the problem in the way that the product team really feels it can be rolled out to customers and better used and more easily used by them. I assume it's a non-linear process. It's not like somebody comes to you with a problem and is, okay, we're going to go look at that. Okay, now we've got an answer. I mean, it's, you're more embedded into the development process than that. Can you just explain that? So we do have, we have a development team in Prague that does work with us. And it's depending on whether we're going, we think we're going to actually build a more, a product with aspects to it like a UI versus just a backend solution. It depends on how, you know, how we've decided we want to proceed with it. So for example, I was talking about inventory intelligence for healthcare. We also had pricing science for distribution. Both of those were built initially with UIs on them and customers could buy those separately. Now that we're, you know, in the cloud suites, that those are both being incorporated into the cloud suite. So we have, going back to where I was talking about our team in Prague, we sometimes build product, sort of a fully encased product working with them. And sometimes we work very closely with the development teams from the various cloud suites. And the product management team is always there to help us figure out sort of the long-term plan and how the different pieces fit together. You know, kind of big picture. You've got AI, right? And then the machine learning, pumping all kinds of data your way. So in historical time frames, it's all pretty new, this confluence, right, in terms of development. Where do you see it, like, 10 years from now, 20 years from now? What's that, what potential is there? We talk about human potential, unlocking human potential. Well, unlock it with that kind of technology. What are we looking at, do you think? You know, I think that's such a fascinating area and area of discussion and sort of thinking, forward thinking. I do believe in sort of this idea of augmented intelligence, and I think Charles was talking a little bit about that this morning, although not in those particular terms. But this idea that computers and machines and technology will actually help us do better and be better and be more productive. You know, so this idea of doing sort of the rote everyday tasks that we no longer have to spend time doing, you know, that'll free us up to think about the bigger problems and hopefully, you know, in my best self wants to say, we'll work on, you know, famine and poverty and all those problems in the world that really need our brains to focus on and work. And then the other interesting part of it is if you think about sort of the concept of singularity and are computers ever going to actually be able to think for themselves, you know, that's sort of another interesting piece when you talk about what's going to happen down the line. Maybe it won't happen in 10 years, maybe it will never happen, but there's definitely a lot of people out there who are well known and sort of tech and science who talk about that and talk about the fears related to that. That's a whole other piece, but it's fascinating to think about, you know, 10 years, 20 years from now, where are we going to be on that spectrum? How do you guys think about bias in AI and in data science? Because, you know, humans express bias, tribalism. There's inherent, that's inherent in human nature. If machines are sort of mimicking humans, how do you deal with that and adjudicate? Yeah, and it's definitely a concern. It's another, there's a lot of writings out there and articles out there right now about bias in machine learning and in AI, and it's definitely a concern. I actually read, so just being aware of it, I think is the first step, right, because as scientists and developers develop these algorithms, going into it consciously knowing that this is something they have to protect against, I think is the first step, for sure. And then, you know, I was just reading an article just recently about another company who is building sort of a bias tracker, so a way to actually monitor your algorithm and identify places where there is perhaps bias coming in. So I do think we'll start to see more of those things. It gets very complicated because when you start talking about deep learning and networks and AI, it's very difficult to actually understand what's going on under the covers, right? It's really hard to get in and say, this is the reason why your AI told you this, that's very hard to do. So it's not going to be an easy process, but I think that we're going to start to see that kind of technology. Well, we heard this morning about some sort of systems that could help my interpretation automate and speed up and minimize the hassle of performance reviews. Yes. And that's the classic example. You know, an assertive woman is called abrasive or aggressive and assertive man is called a great leader. So this is just a classic example of bias. I mentioned Hillary Mason, rock star data scientist happens to be a woman, you happen to be a woman. Your thoughts as a woman in tech and maybe can AI help resolve some of those biases? Yeah. Well, first of all, I want to say, you know, I'm very pleased to work in an organization where we have some very strong leaders who happen to be women. So I mentioned Don Rose, who designed our IAH solution. I mentioned Jill Strange, who runs the talent science organization. Half of my team is women. So particularly inside of sort of the science area inside of INFOR, I've been very pleased with the way we've built out some of that skill set. And I'm also a member, an active member of WIN. So the women's INFOR network is something I'm very involved with. So I meet a lot of people across our organization, a lot of women across our organization who have, you know, just really strong technology supporters, really intelligent sort of go getter type of people. And it's great to see that inside of INFOR. I think there's a lot of work to be done for sure. And you can always find stories, you know, from other, whether it's coming out of Silicon Valley or other places where you hear some really sort of arcane sounding things that are still happening in the industry. And so some of those things, it's, you know, it's disappointing certainly to hear that. But I think Van Jones said something this morning about how, and I liked the way he said it and I'm not going to be able to say it exactly, but he said something along the lines of, like the ground is there, the formation is starting to get us moving in the right direction. And I think, you know, I'm hopeful for the future that we're heading in that way. And I think, you know, again, he sort of said something like once the groundswell starts going in that direction, people will really jump in and we'll see the benefits of being more diverse, whether it's across, you know, having more women or having more people of color or however, you know, however things expand. And that's just going to make us all better and more efficient and more productive. I think that's a great thing. You know, there's a spectrum, right? On the one side of the spectrum, there's intolerable and unacceptable behavior, which is just, this should be zero tolerance in my opinion, there's a passion of ours in theCUBE. The other side of that spectrum is inclusion. And it's a challenge that we have as a small company. And I remember having a conversation earlier this year with an individual and we talked about quotas. And I don't think that's the answer. And her comment was no, that's not the answer. You have to endeavor to reach deeper beyond your existing network, which is hard sometimes for us because you're so busy, you're running around, it's like, okay, it's the convenient thing to do. But you got to peel the onion on that network and actually take the extra time and make it a priority. I mean, your thoughts on that? No, I think that's a good point. I mean, if I think about who my circle of, my circle is, right? And the people that I know and I interact with, you know, if I only reach out to the smallest, the smallest group of people, I'm not getting really out, you know, beyond my initial circle. So I think that's a very good point. And I think that that's, we have to find ways to be more interactive and pull from different areas. I think it's interesting, bring, so coming back to data science for a minute, if you sort of think about the evolution of where we got to, how we got to today, where, you know, now we're really pulling people from science areas and math areas and technology areas and data scientists are coming from lots of places, right? And you don't always have to have a PhD, right? You don't necessarily have to come up through that system to be a good data scientist. And I think to see more of that and really people going beyond, beyond just sort of the traditional circles and the traditional paths to really find people that you wouldn't normally identify, you know, to bring into that path is going to help us just in general be more diverse in our approach. Well, it certainly seems like it's embedded in the company culture. I think the great reason for you to be so optimistic going forward, not only about your job but about the way your company is going about doing your job. What would you advise young people generally who want to crack into the data science field, but specifically women who have clearly are underrepresented in technology? Yeah, so I think we're starting to see more and more women enter the field. Again, it's one of those people know it and so there's less of a, because people are aware of it, there's more, there's more tendency to be more inclusive. But I definitely think, you know, just go for it, right? I mean, just if it's something you're interested in and you want to try it out, you know, go to a coding camp and take a science class and there's so many online resources now. I mean, there's, you know, the massive online courses that you can take. So even if you're hesitant about it, there are ways you can kind of be at home and try it out and see if it's the right thing for you. Just stick your toe in the water. Yes, exactly, exactly. Try it out and see and then decide if it's the right thing for you. But I think there's lots of different ways to sort of check it out. Again, you can take a course, you can actually get a degree. There's a wide range of things that you can do to kind of experiment with it and then find out if it's right for you. If you're not happy with the hiring opportunities out there, just start a company. That's my advice. That's right. I agree, I definitely agree. We appreciate the time and great advice too. Thank you so much. That's absolutely great. Lee Martin joining us here. At Inform 18, we're live in Washington, D.C. You're watching the exclusive coverage right here on theCUBE.