Data Analytics & Decision Making - Amaresh Tripathy

Speaker A: The goal of Data Transformers podcast is to accelerate digital transformation by bridging the gap between business outcomes and rapidly advancing technologies. And we aim to bridge this gap by focusing on data. I am Peggy Tsai, top 50 women in tech influencer, co author of the AI book and data governance expert. I'm Ramesh Danta, an entrepreneur, a tech blogger, and AI enthusiast. Hello everyone. Welcome to one more episode of the Data Transformers podcast. And for today, I'm very excited to introduce a guest who's, uh, been in the data and analytics industry for quite some time. So, Amaresh Tripathi, he's a senior vice president and global business leader at genpact. Amrish, I'm honored to talk to you. Welcome to the show.

Speaker B: Thank you. Thank you, Ramesh. And, uh, great to be here.

Speaker A: Excellent. So to get started, Amrish. So, uh, just why don't you introduce yourself what you do at genpact?

Speaker B: Uh, yeah. So, Ramesh, I run the, uh, analytics, data analytics business at genpact, uh, which is essentially means, uh, everything from our technical perspective, everything on the data, everything around insights, and everything around our AI services, uh, that we go and take for our clients, uh, which fundamentally are our entire philosophy being making tech work for businesses. So that's kind of the intersection we play in and then how to drive, uh, data and analytics and better decision making through data analytics in large enterprises. That's kind of where I focus on.

Speaker A: Interesting. So when I looked at your profile, something that really came across to me is that you are ahead of 8,000 plus data scientists, data engineers. And I said, oh my God. So I mean, like they're only 24 hours in a day. So, uh, how does he manage such a large organization of data scientists and data engineers? So, um, what kind of challenges, uh, do you face in managing such a large, uh, organization?

Speaker B: Yeah, one is more than 10,000 right now, so that the profile is a little bit old and the market is crazy. Uh, the challenges, I mean, so there are one set of challenges that you would think of managing any large organization, which is, uh, uh, how do you kind of communicate one vision? How do you have, how do you have. Once you have a vision around where you want to take the business and the practice, how, how do you communicate that? So everyone kind of aligns to that or have, has an understanding of that and they can articulate in their own words. Right. I mean, it's harder in any capacity as any leader is harder to kind of get that alignment. Our large, large group, which is spread over 30 countries and different time zones, even Harder. Uh, so that's obviously one thing and which is a lot of our management is via large scale communications.

Speaker A: Right.

Speaker B: I mean that's one mechanism to do that. Uh, second is there are ah, uh, challenges around managing just this particular kind of business that you would have, which is a very high growth, ever changing business. Which is anything that is from a technology perspective. Technology plays a big role in this. So from a technology perspective anything that is relevant here 24 months from now, it is probably not going to be as relevant. Right. Or it will be less relevant. There'll be something new that you have to be kind of.

Speaker A: You're being too gracious. You said 24 months. You said like maybe you meant to say 24 hours. Probably.

Speaker B: Yeah, probably. I hope we don't get to the 24 hour cycle here. Then we'll be like journalists with news cycles. So that's kind of the, that's, that's the other aspect uh, of it. But the third is also like kind of the more, more fundamentally is how do you actually uh, I mean and I speak a lot about it, data science and uh, analytics is an area where there is a lot of hype also. Right. Uh, and how do you actually keep it real? Uh, how do you actually keep it real for your teams? How do you keep it real for your clients? Uh, in the sense of the importance of understanding business and understanding domain and getting depth there, equally as important as getting the depth on the technology side. So I think it's a little bit about managing the team via communication, thinking about kind of the cycles and what would be relevant not only now, a little bit later and kind of getting people aligned to that, but also more fundamentally how to actually keep it real in a very high of market, in a very hot market. Which is some fantastic positives by the way, don't get me wrong. But it also means like you have to kind of create a business that sustains itself over a longer period of time. Economic uh, cycles go up and down.

Speaker A: Yeah, I agree with you 100% and especially again this is my bias. Um, I come from a business side and then uh, you definitely want, you're passionate about making the tech work for the business. Okay, so uh, and then you are very passionate about data and analytics from everything that I've seen. So the challenge uh, managing uh, a large technical organization is that they're focused on technology and uh, whereas you are focused on making the technology work for the business. So um, how do you uh, inculcate that sense of hey, you know, we are not Making the tech work to make sure the tech stack works. But you have to make it valuable so it achieves some the business goals. So how do you do that?

Speaker B: Yeah, I mean, listen, I'm very fortunate from uh, one perspective on joint pact here is that it's kind of part of the DNA. So there is something about the DNA and then I'll tell you, obviously that doesn't solve all the problem, but there's something on the DNA which is uh, a company that started with General Electric and understanding business processes and understanding the value of an outcome of the business and the tools are different. What doesn't matter to some extent.

Speaker A: Right.

Speaker B: That DNA is there. Right. So in some ways when you work in this environment, you always think about outcomes, you always think about business. And that's how we kind of organize around, uh, we have organized a lot of our businesses, what we call through our service lines, which I basically think about, uh, your functional domain areas like supply chain, finance, marketing, sales. If you do that risk automatically there is a business inculcation that happens that there is something about the DNA, there is something about how we organize ourselves that helps. That's kind of why I think is one. Uh, the second piece of it is we have this whole mantra which we call bilingual talent. And by bilingual talent we means your technology talent is important, your business talent is equally as important and vice versa. There are people who come from a lot of business backgrounds who pick up technology along the way and become very good data analysts. And you can start being a data engineer, data scientist and kind of pick up the domain aspect of it. But as a culture and as a focus area for the practice, we focus on making people bilingual as we call, as we call it. So that automatically drives the empathy because I mean fundamentally it's all about uh, having an empathy from a business perspective, from our front on technology and technology perspective of business. And that empathy is what we try to balance by like kind of again as I said, managing through communication, rolling out the importance of this whole tech for business, bilingual talent, how we organize ourselves. It's a series of things. It's not one or two things to do, it's a series of things you will have to do. To say this is we are uh, which may not be always the most hyped up thing in the industry right now, but that is who we are, that's how we're at a core and that actually works for our clients and works for us.

Speaker A: That's going to look like, I mean actually I think at the end of the day if you don't make it work, then we have a bigger problem. Uh, then by the way, I can also speak from experience that even if uh, somebody is an engineer, a technology person, they want to see their work reflected in real world. Right. I worked in a company where engineer worked through multiple years, uh three to five years and they haven't seen a product come to market and they get disillusioned uh, by that kind of stuff.

Speaker B: Absolutely. I mean I think everyone, everyone, you're 100% right. And to be, to be candid, that is what, it's not only the sophistication of what you are working on that always matters, it's the outcome of what matters. And I think there are some things where there's very sophisticated and the outcome is fantastic. So there's a whole balance of things that you have to go and try but. Absolutely right. If there is the outcome drives what we find is a lot of people, ah, outcome drives it and they're skill set on what they want to learn and everything depends on when they know they'll be more relevant to people, to our clients, uh, to their colleagues. If they understand certain things it's more likely they will pick it up.

Speaker A: Excellent. So actually I'll go a little bit more uh, later on into the areas of the employee motivation a little bit later especially in data science. So the one area that I do really want to now um, talk about is um, we are now at the tail end of 2021 and uh, so coming out of pandemic and people are saying it's going to be endemic. So however that's going to be. But this is the time people will come with predictions and um, I have seen that uh, Amarish Sripathi has also started uh, coming up with the predictions. And I read one prediction, that recent LinkedIn uh, post that you read which is uh, very illuminating. So one of the key points that you're talking about going forward in the area of data and analytics industry see this backend and front end. Right. So analytics used to be somebody's IT job they will do. We tell them this is what it is, they come with a dashboard and then we say no, that's not what we want all that stuff. Right. So those days are over. Right. So what you're saying those days are over? Increasingly analytics will be the front end. It will even uh, I mean of course we have even seen the self service analytics where the business people themselves are doing the analytics. Right. So can you talk a little bit about this prediction where you're talking about it's going to be the front end. What do you mean by that?

Speaker B: Yeah, yeah. I mean, see, I mean for folks like me who have had the benefit and the fortune of kind of working in this industry for a very long period of time where I mean actually I was in an. I started doing this work in an era where analytics was not even the backend with it, it was just the backend of somewhere. Uh, here's I think actually so very personally I actually do not think of myself as an analytics professional. I see. But I always think of myself as a student of decision making. Uh, and that lens is essentially what I kind of take to the businesses I lead and uh, to the people I lead and kind of the work I do. And if you see from that lens, uh, what fundamentally is happening is how do you, how do you essentially provide, uh, how do you actually inject data and sophistication of algorithms within the data.

Speaker A: Mhm.

Speaker B: Along with the data to aid your decision making? Fundamentally that's kind of what we are doing it. So we are. And a lot of it is how we are trying to make it faster. So things that took time to organize this and that, uh, and things. I mean think about Excel. I mean that's like a major change in how kind of decision making used to be made. Excel provided a tool of kind of crunching data and organizing it and probably doing some more analysis. Simple sort if function that function it changed decision making because kind of series of things that it was enabling to do so and at the root of it, that is the same problem we are trying to solve. We are trying to basically solve our gut. It's not solved. I mean we make decisions every day with or without the data, with our gut, with our experience by whatever flipping a coin, whatever you use, you make a decision. Decision doesn't wait. How do you make it more effective incrementally. That's that I see is all the game is all about that right. It's almost like it's a game of the blackjack being happening and you are counting cards every time it's a little bit better. Uh, that is that entire cycle is as ah, it plays out. What has happened is there's a whole host of technology that has kind of come in and before that like algorithms that have come in. And like any sophisticated technology, like for instance, I like think about cars right now, electric cars. You open it up, there's no parts to even do anything with it. It's all subsumed. It's all there. You can't really integrate it out. You don't understand the details behind it. So any technology over a period of time essentially becomes invisible. Right. And that is fundamentally what is going to happen. Uh, like, and in consumer things, you're already seeing consumer AI is invisible. You are doing AI on your cd, this and that and everything. It's invisible to you. You're actually doing fairly sophisticated work already there. I think the same thing on an enterprise. Uh, a lot of the technology will become invisible. By becoming invisible, it will just become. Come in front and they want to be like, oh, I'm a data scientist here, you're a business person. This and that. It will fundamentally be about, uh, decision making. And obviously this is a longer cycle that will probably take in enterprises because of various reasons. But that's kind of where when I say it will be in the front means it will become invisible. And the person who is the person who is the data scientist, who is the business person, and they're going to make better decisions because it will become a lot of the technology will become sophisticated enough, fast enough, and invisible enough that it won't matter.

Speaker A: So that's an interesting perspective. I mean, to some extent it's accepted perspective, I would say. It's not that it's something that's, uh, out of the loop. Right. That's fine. So, um, on the other hand, I also want to just get into this other area of where you talk about it. Uh, the data and analytics is like playing a blackjack. Right. So one of the articles that you read, you wrote about is that. And some of the key points that you make, um, in that one are, uh, like, for example, it's a long game. Right. It's not just don't look at something, uh, that you can just do in a short term and then done with it. And then other points like, you know, they're not break the silos and. Right. Um, here's the point, Amaresh. So my challenge with those kinds of things is that in practice it's difficult. Right. And especially in the days when you have a chief data officer whose supposedly average tenure is less than couple of years or something like that. Right. So you're forced to. It's almost like a football coach. Right. So you have to deliver the results sooner than later. So. And then break the silos. The silos continue to be there. Each organization wants to make their own impact. So in practice. So, uh, how do you think you can go and reach that vision of, you know, hey, you know, um, it's a long game. So. But play for the long game and break the silos and things like that.

Speaker B: Yeah, I mean obviously you're 100% right. It's very good to have an ideal vision but on a day to day basis you're going to operate towards an ideal vision. And how do you kind of do that? But I think the point is as I said, making it less about data and less about analytics and more about decisions. That's I think number one because uh, the silos and everything and all of that, it's much easier to align people on decisions rather than kind of my data, your data, your rights, my analysis, your analysis. That's uh, on a day to day basis you want we can do that. Second is uh, I think there's on a very tactical basis when you talk about your data and everything, think about where you some decisions and if you look at from that lens there are some decisions that need speed that some. And some decisions get a lot more value from sophistication.

Speaker A: Okay.

Speaker B: There are some that like for instance if there is uh, uh, like a fraud uh request for you. Speed is both but speed actually matters a lot more. Sophistication in some cases like yeah, I mean your cancer detection and everything. Sophistication matters a lot. If you're doing an image, image classification there, I mean you better be sure there. So I think just thinking about what decisions, what tools and how in a very pragmatic way and how you can kind of do it and not everything requires a very, very deep machine learning algorithm. Not everything requires my uh, Excel is good enough in a lot of these cases. How can I make progress on better decision making in the organization? I think there are many ways to kind of do that. But that gives you a right to kind of start breaking down silos. That gives you the right and the credibility to start breaking down uh, kind of playing the longer game slowly, slowly. Because in some ways like blackjack, the main thing you can only play in the longer game if you know how much you want to bet in each hand. Otherwise you would get wiped out before you, before your strategy falls through. So how much you bet and how you actually play that game that actually is even equally important obviously.

Speaker A: So win each game, uh, so you can walk out uh, over the long

Speaker B: term with you know when to bet, how much. That's basically an equally important uh skill than just getting the right on the long run.

Speaker A: Yeah. So excellent, excellent. So and with that uh, the beginning you did talk about uh, some of the you know, emerging technologies that are accelerating at a rapid pace. And then uh, in the passing you mentioned something called artificial intelligence. So yeah, I did hear about it. So for the last couple uh, of years it just been like a crazy in terms of uh, you cannot have any conversation without mentioning artificial intelligence and machine learning and all this. Okay, so with that, uh, what are you seeing in the marketplace with respect to um, the projects uh, that require a mix of things like AI and ML or is it just that that's a new shiny object that you have to somehow test the waters, you have to find a project to use the technology, otherwise uh, you will not be relevant. Is there that kind of a pressure or ah, do you think there are meaningful projects that can really use aiml? Are we at that stage?

Speaker B: Listen, it's both, uh, it's both. And it's all about, I mean being an analytics person, you think about distribution, right? Uh, like yeah, 20% of the time or 15, 20% of the time. You, the problem is very ripe for that kind of sophistication and the right tools and the right data is available. Absolutely. You have to go and use it. Right. I mean we do work for instance, uh, around predictive maintenance around when kind of planes land and they kind of spit out like a bunch of data from their engines. Uh, there is a you on a real time basis. You have to very quickly identify which specific part needs to be proactively repaired and kind of turn the supply chain on a real time basis so that you have the plane and the part meet at the right place and you kind of do the right thing with them and avoid flight delays. Right. It is something against like massive volumes of data is available. Uh, it is where speed and sophistication matters because you want to like, you want to essentially make sure the right parts are kind of, you're doing the maintenance at the right time and there's a, there's a value case around it, you can create a value case around it. Great, right. There are a lot of other cases where, listen, I, I can provide you uh, the right report or the right information at speed, on a real time basis with speed, with very little analysis, some filtering. That's good enough. Right. So I think it's. So there's always a spectrum of 20% probably is there 80% kind of is which will require.

Speaker A: Mhm.

Speaker B: Like not as, maybe not as AI sophisticated. That's like level sophistication or like deep learning kind of algorithm things. But on the other hand then the other problem we have is the definition of AI is also very broad. Like a logistic regression is actually technically an AI algorithm is a classification. And uh, uh, algorithm K means is a classification of AI algorithm. So how you define it also kind of you have liberty to kind of play. AI right now is a marketing term. Yeah. It's not a technical term by and large in the industry. And that's fine. Uh, the philosophy behind it, which is kind of the other AI which I'm really interested in, which is what I call augmented intelligence.

Speaker A: Right.

Speaker B: Which is it is not about AI as in. Because the whole vision of AI and kind of the Hollywood vision of AI is like say self driving cars actually have kind of reinforced it that it is all autonomous decision making. No, most of the decision making in enterprises is actually augmented. It's like people are going to make the judgment calls. People are, I mean the salesperson is actually going to decide, the doctor is going to decide whether they're going to. It's a cancer, cancer cell or not. The uh, credit collection agent is going to decide how to actually make the call, whether to accept the machine's recommendation or not. So it's actually a confluence of man and machine working together to make better decisions. So with that expense, AI is everywhere. That augmented intelligence as an AI is everywhere and it should be a lot in a lot more places. So there's a couple of ways to cut at it. One is a marketing term. Second is which ones truly require sophistication. And third is like automated intelligence or autonomous intelligence or augmented intelligence. Whichever way you kind of think about it, uh, you'll get to a slightly different version of it.

Speaker A: I actually like that the way, the way you're putting it. Just think of it as stepping stone. Like it could be automated intelligence. So I started with the automation. Right.

Speaker B: So that assisted, I would say assisted intelligence. Augmented intelligence. That it could be autonomous intelligence.

Speaker A: Yeah, exactly. Okay, so take those, um, in those areas. Okay, good. So a little bit of segue into a different topic, uh, before I start talking about your journey. And as if, um, you didn't have enough, uh, things to do with managing 10,000 plus, uh, workforce, a data scientist workforce, I should say. Uh, you are an adjunct professor at um, UNC Charlotte. Uh, you're also founder and chair of board of, for School of Data Science. Yeah. So, and then you're still continuing. I thought maybe you started it and then got too much, you got out of it. But it looks like you're one of those people, you know, you really loves a lot of pain. So what's going on Here?

Speaker B: No, uh, I've always been, uh, so I haven't. I've always wanted to be an academic and obviously just in terms of life and you kind of, you end up where you end up. Um, but, uh, so there's a love, there's a passion for teaching, uh, and a passion for getting involved in research and academics has always been there. Uh, which is kind of reflects in kind of the work that I've been doing with University of North Carolina and uh, at Charlotte. Uh, and it's really fascinating because it's a program that kind of I got fortunate to be involved in during a fairly nascent stage where they had a small certificate program that it became like a professional master's program, then like a big master's program. Now it's an undergrad program with a PhD program and it's School of Data Science. So, uh, it's been very gratifying to kind of see over the years as the field has developed what is academics, kind of what the academic world kind of has caught up with it. And one part of it is kind of me, in my own way giving back. Because I see where the industry is and I see where the academic is and kind of where the two meet. That's my way of kind of giving back to the world of academics. To be honest, I learn equally as much in return as to where the future is. I told you about this 24 months where things are going and everything I get, I get uh, like a, uh, bird's eye view of, uh. Okay, what some of these things are coming up. Uh, what is the talent want? How do they are thinking about it? What's the mix and profile of people coming in that gives me, makes me a much better business person, on the other hand. So it kind of works out pretty well.

Speaker A: Excellent. So actually, um, same thing. I work extensively with the University of California, Irvine. Um, there's a different vibe that comes from working in that kind of environment. Excuse me. And then, um, so, uh, we think we are giving, um, you know, so much to it, but we are getting a lot more. At least I'm getting a lot more

Speaker B: out of that exact same year.

Speaker A: Thank you for listening to today's episode. If you liked what you heard today and would like to hear more, please subscribe to our podcast on your favorite player like itunes and Spotify. And please do rate our podcast also please go to our website, www.datatransformerspodcast.com for more episodes, blogs and information on our speakers. Thank. M you, Sam.