Hi, very good evening to all of you. And thank you so much for joining us today. You know, this is our regular August, this August edition of our industry experts series, where we get senior professionals in the industry to talk to our students as well as prospective students about interesting areas in, in deep tech and technology, etc. A very contemporary topic today and a very interesting person joining us, you know, and we are going to be having a very interesting session, I am looking forward to it, I know, you know, the speaker has, is doing fantastic work in the space that he is in. So, and I'm going to be introducing him in a little bit. And you know, I'm going to hand it over to you, to him. So what we have today, what we are going to be doing is there will be he'll be sharing his thoughts, you know, in around data science in the post COVID world, you know, very indicates it's something which is very contemporary, it's something that a lot of us will have a lot of takeaways from. And we will also be taking questions from the audience like we do for every session that we do. So what I will request for everybody is to hold on to your questions, you know, till the end of the presentation, and then you can send your questions across in the chat window. So and we will take as many questions as possible. That's there. Now, it's my pleasure today to welcome Peter bit hung over with us, we have bitten Roy, he is a seasoned professional with about 13 years of experience in financial services and in healthcare AI. He currently is the director and the AI Center of Excellence leader at Fidelity Investments in India. You know, he has worked in multiple functional areas, such as customer experiences, operations, risk mitigation, and new products. Prior to his current role, bihan worked across multiple data science and analytical positions, and building analytics and building the analytics and AI practice for call centers back offices and client service management roles. So behalf over to you. I you know, welcome to the session today. Firstly, you know, thank you for taking time out of your busy schedule. And you're having this conversation with us. I look forward, you know, to the insights that you will be sharing with us today.

Thanks. Thanks for retro and thanks. Thanks to you and the talentsprint. I think, you know, my pleasure to join all of you. And, you know, definitely look forward to the next hour or so I think, you know, looking forward to sharing some quick thoughts, I think a lot of the thoughts that I'll be sharing will be a reinforcement of the thoughts that may have been read or shared. But I think it's a, it's an interesting topic and an ever growing field. So look forward to the questions. As we talk through the topic of data science, I will probably start with, you know, my round of introduction. I know I throw Thanks for the introduction, it covers quite a bit, but we'll give a brief which which will help probably, you know, the folks who have joined us, you know, how the career looked like. And now we can think from that one of the dimensions. And then I'll start with like, when you know, when talentsprint and I know I was working with Rama, you know, she gave you all came up with this topic, data science in post COVID world, I'll start with a very interesting thing which will clarify the post COVID world so that, you know, we I know, there's a lot of questions on pre and post. So we'll start there. So now getting started. I will say, you know, I have been with the industry in data analytics, Ai, you know, over over 13 years now, I started, you know, way back when data science work word as such was just getting started. Or it didn't exist in a formalized fashion. To be honest, it was more around statisticians, decision scientists, analytics, you know, data analysts, business analysts. So I joined, you know, what used to be referred as business analytics groups and business analytics and research team within Fidelity Investments in started my career and have always been part of in the data analytics industry. You know, started with very early on, you know, doing analysis, business analysts is prediction, kind of models with, with with, with data that's available, not the big data that we talked about today. But over the years, what I have learned is we have new new jargons new words that have come to the industry forefront. However, the premises remain the constant, you know, they the focus is on insights, the focus is on the value that the CEO gets out of the investment in a data centric world. The focus is on how do we build smarter products? The focus is on how can AI help humanity? How can AI make a difference in the society? So I'll definitely say data science is an ever growing field. It's, it has its own pros and cons as well, I talk of I'll talk about it. However, the journey has been exciting. I don't regret joining this industry, I am passionate. And I think this will be my focus area over my remaining part of my career as well, because it's always exciting to find how data and science and analytics can solve problems, which earlier was thought of cannot be solved. So right now, you know, as you mentioned, I do lead the group of, you know, very talented data scientists. And, you know, it's it's very exciting to see how we come up with, you know, patent patent pending work streams, a lot of value in the business and an ever growing new areas of work. So, with that, what I would do is I get started, and, you know, we'll jump to the topic right, right away.

Let me go to the first slide. The first slide would be probably the most, I don't know if all of you have seen this slide. If not, then I will definitely say this is the first time you're seeing but I think this is a very common slide put it in many places, is it's the data in data science interest over time. And the word interest, you know, to start with is the Google Search interest for the keyword data science over past 18 years. Now, you can see it's like it existed in 2004. As I mentioned, you know, there were various facets of thing. And, you know, in the last two decades, you know, in the last two decades, data scientists word didn't exist, probably not it existed in a very informal fashion, in different pockets in academia. But in the mainstream commercial application, I think it has been more over the last decade, which has seen a breakthrough. Some people definitely claim the data, they organized it analyze the information. But the data science professionals we admire today, you know, stand at the head of a relatively new and vaunted career path, right? The modern days, data scientists, they have skills that merge the technical know how, and how an analytical expert with curiosity, you know, they kind of solve the problems, you know, that's the exciting part of today. The interesting part is this word data science. And the data scientists, as a lot of people have said, it's a part mathematician, is a part computer scientists, and, you know, part trend spotter, you know, some of it's been, it's written in one of the SAS insights, you know, and because the straddle, you know, both business and it, they are kind of highly sought after, in the world, through this chart, what I wanted to start with was know, to clearly call out that the interest, you know, the 100 means, you know, value of hundreds, it's like the peak of popularity, and I think the value of 100 was there, you know, whenever you looked at a point in time, it was always there. And as you can see, it did pick up again, you know, and this is the period, which we refer to as COVID. As, when COVID hit the world, we are still in COVID, we're not out of it. But what I wanted to show is this job and a lot of the questions I've seen with as I talk to folks in the industry, and as I, you know, in practice and other exchanges, it's kind of, you know, why why are we seeing, obviously, a dip in the demand? Are we seeing a dip in the need for a data scientist? My answer would be, you know, let's go back and take a look to this period of the, you know, the term where if you look at them starting to spike up, and then we did see some, you know, what I would call, you know, cold, cold kind of nature from an overall demand and data and analytics, or data science kind of thing. And the reason was, you know, 1213 years back in 2008 789, there was a, there was a market price of the lemon brother crisis and other crisis. So it was kind of there were a lot more interesting and lot more challenging times for the world than to kind of look at data science as a topic. However, data science, and the decision science never stopped is just an interesting science. If we look back at that, and we look at look back at a period, which was just right now and just last year, I would say it's, it's nothing called Data Science in the post COVID world, data science, which is making sense out of data is always in the increase, and it's only going to increase. So I would like to start this entire session saying that there is nothing called Data Science post COVID it's only growing the data is growing the need for data science, scientist is growing, the need for making sense of data is growing. So I would just say there was a recent article, you know, in the, in the India today, and that article clearly said the career outlook for data scientists in India sky high pay and rising demand. So the demand Our data scientist in India is at an all time high, which is resulting in an upward career arc and generous pay from companies. And this is like as, as recent as last month, and this similar topic has always been over the last several years. And it just continues to be that way. So I will say, data science in the post COVID is, is just a myth, it's just what I will say is the value that a company gets from data is only increasing, and they want to do more. So that's the point of view, I will say in this entire topic.

As I move across to, you know, none of the none of the next one of the next topics in I wanted to, interestingly, so when you when we look at this, I had an interesting twist. It's almost like my inquisitive thinking as a data scientist, like, why do we see this increase in, you know, this trend? You know, and I had an interesting take on it. It's funny. It's like, why do we see that graph increasing? The graph is increasing, because, you know, a few years back long back, you know, nobody cared for math geek in parties. And right now, people love data scientists and mathematicians, you know, we are not no longer calling them math geeks, we're calling them data scientists. Right. And because we last have started loving them, we just have this graph exponentially rising, and the demand and everything is growing. And just on the on a lighter note, I think that's how, you know, the rise of data scientist is connected to the graph, and this is a beautiful maymay that I have found it. Continuing on this, I'll definitely add, you know, one more thing is, how many of us have seen this? In this topic, right? I think we all have seen in most companies that we've worked for our most companies that are talking outside, they always talked about, I want to achieve, you know, achieve our objectives of the company, you know, company x, x and replace x x with any company by 2021 234 2030 2040. Boost organizational performance at all levels, use an integrated, smart digital system that can overcome challenges and provide quick, efficient solutions, make company xx, the first field first in the field of AI investments in various sectors, invest in the talent and skills of the future. This slide is important. And this is helpful for all of us to learn that these objectives we can replace with any company, and probably this will be more or less true for most companies not start and there will be changes how we can achieve this. What is our what is the CEO of the companies? And what our Chief Data Analytics chief data scientist? What are they expecting of their data scientists, they are expecting that these objectives are closely aligned to the artificial intelligence objectives, and how can my data AI can help me achieve this. So what by this I meant to say is, this is where the focus has come. Now, I think the focus is, Hey, I have all these objectives. And I believe all these objectives can be achieved by use of data in a smart way, in a more meaningful way, in a way where I understand my customer better in a way where I can engage my customer firsthand in a way where I can serve my customer, even when they are not thinking of me. So I feel like this is more towards, you know, helping that, you know, that objective meet through the use of data science. So these objectives, more or less remained the same. What is what is true to this is our industry and our field, which is data science can help achieve this larger objectives of the company by use of data, and data science and AI and machine learning all the different terms that we hear about. Just for everyone, I think what I'll say is, you know, put all your questions out there. I think as we go through the slides, at some point, when I'm done with all the slides, we'll take all the questions together so that we can address it at a holistic level, so that I can flow get the complete flow. So you know, keep putting in the questions if you have any of them. As I continue, you know, the most, the big question that we all of us have is what do companies expect from data science, that data scientist, right? We just talked about what the goals that data centers help achieve. But you know, when I'm searching for a job, or when somebody is looking out for a job, what what do the Chief Science Officer, our data science managers, or the business head? What do they expect? From the data scientist? Right, and I think there is a, there's a good take, and I think, you know, this is this is sourced, you know, from, from one of the places that I read about Intel, Pat, I think had put it nice way, is like, what what do data scientists do? You know, they find data analytics issues, like you know, which can be solved. Let's find those issues. find meaningful information from the big data, like, you know, just start finding whatever data we have. But we know that the data scientists, they're good, very good at asking questions. They're very, they should be very good at knowing, knowing the trends and asking like, why this is happening, or what should I look at or you know, these kind of questions and it's inquisitive nature which will help also evaluate the third thing, which is evaluate new data, or the new data sources be both internal and external, right? It's not just restricted nowadays, to your company data or the data that you have available. There's a lot of social data. There's a lot of Linden later, there's social media data, there is Facebook, there is, you know, there's lots and lots of unstructured data that has that has come to the forefront.

The next phase is how can we support this data collection, integration, retention? You know, all of these, these are like, very important, as a part of data governance as a part of usage of data. So I feel like these, it's very important. As you can see, almost the first half, there is a significant 50% is identifying the problem and understanding more data I have, and what data can I use, and how can I really use it in a legitimate manner with privacy, and a lot of those things are also coming up. So that's a very important part that 50% and we are still not talked about what most people are excited about is the garden. The know, the fifth part is understanding the problem. And then classifying the problem in a right way, like use suitable algorithms to solve the specific business problem that has been identified. Once the algorithm is done, which probably contributes to some meaningful time, but a lot of times, that may not be just the big time. And that's a very important thing in a data science. What companies expect is, I'm not looking at another algorithm, I'm not just looking at another potential modeling technique, what I'm looking at is how can I really solve the business problem, which is the issue at hand or the opportunity at hand? Once we have the algorithm and the data, the goal is like, Can we now have the algorithm identify the target population? Or the you know, the risk areas identified? Or some new opportunities identified? Now basis that how can we craft experiments to support the assumptions that we started with, which is the experimentation so experimentation and AI machine learning go very hand in hand, we have new ways of machine learning techniques, new ways of doing things with data, go, Okay, today, everybody is okay to do whatever they were doing 90% of the time, they are okay to experiment and percent of the population with new techniques, which helps us advance in the in the field of data science, it helps us only advance more as we move forward. The last is, which is very important role. And I think it's a skill as well, is a lot of times we as data scientists, and I have done it myself as well. In the past, I'm still learning how do we not just think of data, data science, data collection, the algorithm, the type of problem, deep learning, machine learning, but really collaborate with stakeholders to kind of identify the business needs, you know, what did the business really want, you know, that is very important, we may have the best model, but the best model may not be of relevance to my stakeholders. So I think collaboration, communication is a very critical component for the success of the data science project. And there is an evolving role that is coming up. And I would call that almost like a business translator, somebody who understands the business, somebody understands technologic enough, somebody who understands data science enough. And they are the translators who are bridging the gap between the technical geeks to the business savvy people. So that's a role which is which is growing, growing a lot. And I think that's something which can only happen over time, we need the understanding of data, we need understanding of technology, we need the understanding of science, and we need to be able to understand and connect with the business and the domain that we are working on. So I said this is at a broad level, I thought a very, very simple and very good way to look at how an AI project could could be thought about any any analytics project to be done. Or to be clear, this doesn't get into technicalities, just it talks about how what a CEO or water chief company officer or chief data officer is looking for while they're solving the project, and they will rarely look at status updates or learning through the project. And where is the bottleneck, lot of projects today do not see the end of life. Because it gets stuck either in the data part, or in the experimentation part. Or even in like, Hey, you know what, I'm not comfortable doing this, which means we never communicated with the stakeholders, like whether they're interested in doing something like this, you know, did the design and everything. So that's very important. And this gives us a good layout from a thinking perspective. Just to keep it in a little light. I think at the end, I just talked about machine learning or you know, our analytics technique like an interesting take, like, you know, I just picked up you can read it. I just read for everyone, this is the machine learning system. What is it? You poured the data into this big pile of linear algebra? then collect the answers on the other side? What if the answers are wrong? Just turn the pile hunted start looking right. So I thought like, you know, this is what we all do a lot. And I feel like you know, this, this is this is quite interesting as we thought about it.

I'll take this you know, there is always a question. on, you know, that What does data scientists do? What who is the data scientist, right? And I thought like, you know, this view that Gartner had in a variety of analytics and AI ml roles and skills, we can see it as a continuum, we can see it as independent each, each volume is independent, whatever way we want to choose my takers, we can pick up if we are in any of these columns, or we are not in the columns, it's like we are we are just getting out of college or you know, planning to learn AI. You know, these are the skills required at a broad level, I felt like this was a very practical way of looking at different types of data scientists, the term data scientist is being used rather loosely in a lot of times, it's being used loosely, from analysts to data visualizers and business intelligence experts all are being termed as data scientist, while this loose net of a definition is not entirely wrong, a data scientist can basically be defined as a person who is part mathematician, like I mentioned, the beginning part, computer scientists, and a part business transporter was able to understand the business was able to understand and connect the data and the business and the it and the both the worlds. So data science is actually now being integrated with industries across all sectors. And that is why like the data scientist, you know, are expected to have a broad set of skills. But you know, what are employers expecting? The employers are actually expecting in an agile framework? How can all of these different skills come together and work and provide value to the business means every individual will have their own specialization, but there will be a crossover? How can we collaborate to take it forward? The concept of a full stack data scientist? I know there may be there are few who are very good, full stack data scientists. But is it really a true way to look at how the data science world? My take on it so far is no, because it requires a very different dimension, when we start thinking, the previous picture we talked about, like very different skill sets very different things. And I think it's a big enough word. And that's one of the things I talk at the end, why it's a con like it's an ever growing field, and a specialization is what is needed. And I think what this picture talks about a quick call out, like, you know, these are the various skills at various level, like data knowledge is similar to the data in that domain knowledge where you belong to, it's very important, what's the data in there and what's not there, if you want to be, you know, either a data analyst or a data scientist, you know, that's important. The data skills, business acumen, ml modeling skills, coding skills, soft skills, soft skills means the, you know, visualization to lot to some extent, telling a story storytelling is very important part of data science, and communication with stakeholders are they're all part of soft skills, you know, you know, what kind of things we need to solve, what can we really solve for we are very important. And, you know, these are the analytics process skills, which are expected from each of these groups. So this is a very fairly practical way of looking at the data science and different continuum of analytics roles. If you look at business analysts, which is what I said, when I started long back, I think we did not have a term called our data scientist title. What we had was like business, business analysts, and business analysts person used to, you know, have was forced to learn the business knowledge, because without business knowledge and the data knowledge, there was no way we could get insights, because at the end of the day, insights to business is what matters. So if you look at the soft skills, and business acumen are like, top two, and this means like, I am understanding the data, which is the data skills, and I get those insights using modeling techniques, or, you know, certain exploration techniques and something, and I'm able to convert the business problem into a meaningful statistical problem at that point in time. And it's still relevant today. It's very much needed today. Because today, there is we just don't need coders on one hand, we also don't need people who do not know coders, what we need is both of them. Right? That's where the collaboration comes in. And I think this is a good example, I think a lot of you would probably be in either of these fields, yourself, or you are looking to, you know, be one of them. These all are kind of different types of data science roles. I would like to highlight one of the things which is being taught in the industry, the concept of citizen data scientist, a citizen data scientist, is kind of if you look at very closely what was with the business analyst in the of the past, but a significant call out is they have advances in terms of ml modeling skills and coding skills. If we look at the expectation of the citizen data scientists, they are way more superior when it comes to ml techniques. And they're definitely far better when it comes to coding skills. But they are not the ones who are doing advanced research or their interest is something but citizen data scientists can bring lot more value at a reasonable cost to the company, but they can still generate a lot of insights because they're willing to understand the business.

What is the big difference between a citizen data scientists data scientists, I think very if you look at it's predominantly the coding side, I think that's the big part. Where what is a data scientist and a big difference? Well, the difference is like the business part, if you look at the soft and the communication part of it, that's why I said, like, if you merge, both of them is where the next set of roles come up, as you grow in the company or something, you know, I, I understand the coding, I also understand, you know, the communication, I understand the business well, so can I just, you know, that's the multiplier effect of the jobs and the roles that companies are looking for. What that helps is, it only helps us accelerate the goals of how to achieve those artificial intelligence objectives of the company else, it's a very siloed way. I get a lot of times question, hey, I'm part of a data analyst, or data analyst, I'm a data engineer, I'm an application developer, can I go to data science? The answer is absolutely, yes. The answer is, you know, you could pick up your choice of subjects there are, if you search, like the Google Trends, like you know, data science, career path, or data science, how to learn data science techniques, and ml techniques, you will get plenty like you will get a you know, plenty of roadmaps will get plenty of, you know, Institute's who will do everything to make all of us prepared. So I will say it's just a matter of choice.

Hopefully, this gave us some clarity. This is this, I wanted to have a very interesting, you know, conversation with you like, there are two types of data scientists, like I said, citizen data citizen, data scientist and data scientists, but the way I was as I was exploring this whole part, a couple of things, which came to my mind, it's like, this interesting question, which I talked about, what is data science? You know, and there is a Korra. And there is answer from somebody in Cora, you know, Michael, I read it. And it was, it was interesting type a data scientists and type B data scientists, if you recall, the citizen data scientist, maybe to some extent, so that you we can connect the dots from there to here, I'll say, consider type A as business analyst, or decision scientist, or, you know, are probably business analysts, and type B data scientists or the pure data scientists, which we just talked about, you know, type A, this a typo here stands for the A is for analysis. And type DB here stands for building, right, I picked up this picks up also from the kinds of data scientists that is there in our HBr article, and I feel like it will be it will be interesting for you. And I think it can help you to learn that it's not just coding, it's not just the value is algorithms. It's like, how do we look at a lens that what can we bring value to our, to our business? Right? So I'll say, when we look at this design, it's predominantly analysis, the citizen data scientists, data scientists, they're primarily concerned with making sense of data in a fairly static way. The B type of data scientists, they have similar to large extent, a lot of times, but the main thing is they're very strong coders and maybe trained software engineers. So the a is focused on lot more on insights and modeling and you know, focused on those concepts, which are there in statistics departments, you will see them very strong in statistics. And you know, the the foundational topics and everything. The Type B is are the people who are dealing and want to deal with data in production. This is where the big data the entire world is moving towards, right. And typical problems on a Taipei would include, like designing and analyzing multivariant, tests, research and informed strategy of the company. Whereas type B is like building the recommender systems and improving search a lot of those things, right. But I will take a twist here to say that a lot of times, I hear, hear, you know what, and I know I'm showing you directly here, like all the four or five things together, but a lot of times I get a question, by the time I reach, you know, the type of problems I'm solving, I want to be in the Type B, I don't want to be typing, what I wanted to clarify is type A is as important as type D, because what type A is the consumers of type a type of work, you know, they are consumed by the senior executives, they are consumed by the decision makers, what I would refer on the left hand side my, the header of the slide, this is called the data science for humans, this type of data science, or this type of data scientists actually help the executives, the product managers, and the designers, you know, to kind of identify what decision do I take basis, the findings of the modeling techniques. So I will say data science, these are the consumers to look at it. There can be many types of algorithms, but this helps them draw conclusions because at the end, humans are going to make a decision. It's like we did a marketing campaign, but which yielded higher revenue. Right. So it's like which part of product experience is suboptimal? So I think it's a culmination of those kinds of meaningful business work which still needs a data scientist. The Type B is almost like, what is very important point the consumers of this output the consumers here are executives, the consumer software data scientists or computers themselves, right? So they are like, the examples of this work is like a recommender system, right? The recommender system is, you know, we get the algorithm, the algorithm gets an output. And then output is shown to me as a recommendation that began by this T shirt. Right? Or, you know, which is a good t shirt for me, or what militia what medicine order could somebody prescribed, right. So, I think it's kind of computers, learning versus one is like an output to them. So I thought, like data science for humans and data science for machines is a good way of looking at where we want to head or an interesting way of looking at this type of work. And this is more scaling, this is more insights and eventually scaling.

I could quickly go, you know, go over a few things. Now. Just you know, stepping away like, this is one interesting way of looking at different types of AI applications. Sense comprehend act is similar to what humans are supposed to do, we can do this very good. We can sense things, we can comprehend things, and we can act on things. And the future of AI is this, that's what the expectation of AI is. However, we are still a lot away. But I think at least at a broad level, the types of technologies that AI has can be classified broadly into this when we say sense, it means the machine perceives the world senses like machine, which is to computer vision, audio processing, it's like perceives the world by acquiring and processing images, sound speech, text, and other forms of data. comprehend is a machine is able to recognize patterns and context information in the in the information it collects, just as humans and interpret in the following way by understanding patterns and context in their perception of the data world, right. That's, that's the comprehend part. And the act, what is act like we act on things we know. So act means AI, the type of AI, it enables a machine to take actions in the physical or digital based world, on the perception or the comprehension. Right? So I feel like, it's a very nice way of looking at things. I don't think we can go into a full hour of thinking through and talking through this topic. But I thought a good way to look at this is sense, comprehend Act and the type of solutions, you know, illustrative examples, like virtual agents, we all know, virtual agents are those where, you know, it's it's we think we are interacting with a human. But eventually it's it's a bot and the bot is able to understand the context of the query and everything. I think that's that's those are, those are going big. In banks in, in, in retail e commerce wherever we pick up right. Identity analytics is is the entire world of biometrics. I think the future is computer vision, how can we do face recognition and everything? I think that's a big area with regards to security and everything. Cognitive robotics is an area which is around the Intelligent Automation piece. It's like, how do we get assisted intelligence? It's not artificial, it's a little away. But it's key closure, because you have human robots kind of helping the humans are able to do what humans were at least 100% if not 100% 60 70% of that for speech analytics. We all know how can we understand whatever let's say I'm talking right now get Is there a way to understand what I'm talking about? You know, how can we solve that recommender systems is always out be the recommendation of the products on the e commerce portals, to the emails that we receive to everything there today, it's all recommendation based on the Netflix movies to YouTube. Next, next best song to listen to everything is recommender system, which is kind of machine learning approach. Data visualization is uncovering insights. And I think it's all about how do we tell the story about it?

I quickly go over a few things. I know, I probably spend some time on q&a, these are like at a broad level, you know, basis industry different types of use cases of machine learning, like in manufacturing, you know, the one of the focus areas of machine learning is could we identify, you know, where when does when does a car need maintenance? Right? Could we predict in advance? Could we predict in in advance, you know, what could we predict in advance, you know, what kind of maintenance is required for for this topic. So I will say, you know, this is interesting. In the retail world, we just talked about recommender systems. We have healthcare and life sciences, where I think proactive health management could be area alerts and diagnostic from real time patient data. That's an important area, you know, we are we are all very smart smartwatches. I think that's an important area of healthcare. It's gathering tons and tons of data. I've heard and I've read a few articles where machine learning was able to detect a person is likely to have a needs a heart surgery or his heart conditions are Have you ever read the reports? And I think it's interesting how somebody from a different country helped his, you know, family members, you know, learn that Hey, take take my father take my mom to the hospital. I think it happened in through I think, I don't recall exactly. But I think that's, that's something you can look for which machine learning is a way to look at it. Travel and hospitality, I think they are one of the fields, which have been using machine learning and data analytics for long like predicting demand. dynamic pricing, a lot of those use cases are part of travel and hospitality, financial services, risk analytics, customer segmentation creditworthiness. They have been here over a decade, most of them, what we are doing now is with the advent of data, could we do move away from rules based to more data, machine learning data driven approaches? So it's slowly and steadily moving towards? Can those fixed rules be replaced by more automated rules, because the rules will be learned from the new set of data that we have energy feedstock? I think, you know, could we understand the smart grid management? Could we understand the energy demand and supply supply? Imagine like we could understand the power consumption of different states within the country so that we could eliminate, you know, power outages in a particular state? Because, you know, under supply and get from another state where there is oversupply, right, the the I think there are ways to be solved, but there are a lot of logistics issues, why we can't solve it. So it's more like, how do we take something from from an idea to that's where the focus is? In a sum, I wanted to highlight like some challenges and adoptions of adoption of artificial intelligence, I think lack of enabling data ecosystem, I think most companies today are failing, not just investment in AI, the biggest part of AI, why are we talking about AI? Why are we talking about machine learning, we are talking about it because of data. But data ecosystem is also the reason why we are not able to move as fast as we should be. It appears we have a lot of data, but when when it matters, I think data is still in silos, and the data ecosystem is still still very, it's still it's it's moving, it's moving at a rapid pace. But I think that's one of the adoption challenges. I think it advanced AI is a is a different thing, as you can imagine, you know, cognitive AI, or you know, recommender systems are, you know, the entire smart, smart working algorithms, we need to do a lot more research. So I feel like one of the big areas where all of us can contribute. And that's an very, very nascent and interesting field. It's been there, but I feel the academia, and the industry gap is coming closure. And there's a lot more needs of PhDs, there's a lot more needs of higher degrees. It's it's an area, it's an area of its own, you can make a career out of AI research. So I think that's an area in itself.

I think it's connected to the second it's almost like inadequate availability of the right type of expertise versus general expertise. I think the cost is one of the factors, but I think that's something which companies are aggressively adopting and trying to diverse through it. There are unclear privacy security, you know, ethical regulations, there's, you know, in us, I think, the GDPR in a lot of regulations, you know, what, what does it mean? Can we share data, what kind of things in healthcare, there's a pH I in in the US in the Western world, right? We cannot just share the data and, you know, make decisions and recommend, because it might mean that I didn't want you to see my data to make recommendations to me. So there are a lot of those things, which stops us from really operating at the frontier of AI. So this is almost like, you know, one of the few areas what, what makes this happen? So quickly moving on, you know, I get this question like, you know, data science field and careers, I think I've covered this more or less, but at a high summary, data demand continues to outpace supply, why why should pursue that's the reason. There's abundance of positions, the type of positions, the different types of skills, you just, you just looked at it, right? I think it just, it just immense, you can pick up any field, anything, and it will just cover everything. Data Science is versatile. I think you can move from one one column to other column to other column. And I think it's just you it will keep you hungry, and it will keep you growing in that field, a highly paid career. I will say this, this should not be the goal. But this is one of the reasons I think you can start from very low, but you can go only sky's the limit. I think if you bring in skills, if you bring in new areas of research, that's something why you should pursue. Sometimes I will say what rather than saying cons which is what a lot of times their articles say I will say why am I when in doubt Should I pursue not in pursue and I'll say why you should pursue or why a lot of times we are in doubt, because of certain slides, I shared this lot of ambiguity in terms of what I want to do and what companies look for. It's an ever growing field. Just pick up anything I think will grow. Mastering data science is near to impossible because there's this continuous evolution of it. And the most important thing is there's a need for investment of time and domain. Knowledge, a lot of times we just learn data science algorithms. But the key thing for business value is actually the domain knowledge where we work on. So you have to spend time understanding, if you're in financial services spend the time to understand the business problems, what the CEO wanting from you. It's very different from a healthcare industry. You can switch industries, there are a lot of horizontal skills. But at the end of the day, to make a meaningful impact, it's very important to understand the business that I'm playing in and attend I just covered in other data privacy concerns. In a quick fashion, I'll just cover you know, one second, these are some examples of a research you know, you know, a lot of the work that we are doing on his artificial narrow intelligence, like we are beating the beat go world champions race, read facial expressions, we the AI is able to write music, or diagnose mental disorders, who is likely to have next surgery. It can even comfort earthquake survivors, this is a nice thing I found out I thought I'll share with you. But this is something which the world is looking at, you know, this is the future. But this is not so easy. And this is where I think a lot of thing is still still far. And we're still we're still quite far from there. So I'll just say it's not quantitative, it appears a lot of times quantitative. There's a lot of qualitative aspect as well as a there's a good memory here. Let's get a show of hands, who here prefers quantitative data over qualitative data? It says 123. Everybody raise their hand? Oh, looks like everybody telling me why do you prefer quantitative wins? Why do you prefer quantitative techniques over qualitative, and they start explaining, and then certainly, again, just say numbers, and sometimes not everything is a number. Not everything is just a story.

At the end of the day, it's a, it's a combination of things we need to appreciate. It's not just autism. It's not just data. It's it's somebody. It's a translation of the data science and business knowledge. And that's what comes together as you know, the value to the business. My final slide I'll get there was more like a thank you note. You know, thank you for joining me today. I thought know this David, sign a professor at MIT, you know, he had a very nice, I'll let let all of you read, I think there's a very nice quote, it clearly says what we need to focus on. And what will define success of data science professionals, it relies on becoming trained, and able data scientists with the ability to perform data processing and computation at massive scale. To achieve this professionals must invest time in an ongoing education through institutions with multicolor multidisciplinary query programs that include elements from engineering, Mathematical Sciences, and social sciences. Converting big data into meaningful information begins with skilled professionals who are educated in all disciplines to be both data scientists and statisticians. So I like this definition, which which is a, which is descriptive enough. And this can be a motivation as to how broad this entire thing is. So with that, I think I read through I think, I don't know if if there was anything else you wanted to cover, hopefully I, you know, was able to give some perspectives and know, give some time for people to reflect on.

Absolutely. Thank you so much better. And I think it's, you know, very interesting set of slides, you know, that that you put out in terms of, you know, GDP, the things that you spoke about, we have some questions around when I won't be wasting a lot of time. In terms of in commentary, perhaps we will go directly towards some questions that have been coming in. So let's look at the question by Himanshu. You know, he saw a video of open codecs where engineers showed just by writing simple sentences in English, and Ei system come up with a complete code of the desired program. And there's also auto ml, which is self sufficient in analyzing data. So what will such advances in technology have on the careers of data scientists is something that Himanshu is asking.

I think, thanks, a monster for asking this question. I think it's a it's a, it's a good question. And the way I looked at it is today, and as we move forward, I think there is this concept of in every product, it happens, it happens in every area. It's it's the commercialization of commercialization of anything that is new. And I feel like this is another such thing where data science was not in the hands of everyone today, and in the past, data scientist means as I said, a lot of coding a lot of mathematics. Now, that's not what everybody is meant to do. The value of data science is the insights and the connection to how we drive business outcomes, right. So I feel like the need for data scientists will continue. And we need to keep in mind that auto ml or in a lot of these things, they will have their own journey. And this existed 10 years back we used to have in other products where your analytics like you know, it's a front end, you give the data it throws you predictive analytics, and what we have seen is good for tailored, you know, it's good for variable And regular data, but for the company you're working for, for the division, you're working for the understanding of data and the type of data needed is not done by automatically. It still continues. And I think it will continue to large extent, what may come down is the lot of coding that's required may come down a bit. But I think as I mentioned, one of the challenges is there is this move from narrow AI to general AI, right, I think a lot of the easier stuff will automatically get automated, we will have actually higher order problems like autonomous cars, right, you know, those things, which is still not solved. So we will have higher order problems to solve car, this probably the problems that we're excited about will no longer be exciting humans. So there will be next generation of problems which, which will excite us. And that will, that's the time when it will again, have Okay, now what to do. So I will say it's this is meant to ensure that data science is in the hands of a common person, because the world is changing the software engineers, the data analysts, everybody needs to understand data and make sense out of it. And not everybody is a coder. So how do we give it in those hands? This is why this is happening. I don't see this as a threat. It's more like, you know, augmenting all of us.

Absolutely. Thank you. This next question comes from Jovan. And I think it's kind of a very traditional question, your trade off, you know, a module is very good at sniffing out positive outcomes, but at the cost of high false positives. I think it's a very classical problem that's there. What's your take on you know, this? How do you manage these kinds of models? What What do you know? What are the how do you ensure that the trade offs are at a minimum? And what's the decision point over there?

I'll say, as I mentioned, you know, we talked about, you know, there's this thing, which I was reading, and it's interesting that, you know, is machines for humans or machines for, you know, ai for humans or, you know, ai for, you know, machines, right? data science, for humans, data science permissions, you can call it any way, right? So I feel like a lot of this problems with where we are still not able to do is almost like data science. For humans, it's in the risk world, we have seen a lot of times in the risk world where, you know, the false positives are some of them, they mean, a lot more work. But the cost of missing out any single one is much higher, you know, the reputational risk is a lot of those drops, you know, risk is much higher than probably a lot more work that needs to be done to eliminate those, you know, bunch of not so true positives, right? So what I will say is, this is where if I connect back to my project, lifecycle, that communication with the stakeholders, you know, what's the threshold? What's the true problem they're trying to solve? If we understand that, then the data scientist and all of this can come come into judgment that Okay, I will have a point six times I will have a point seven cut off? No, no, you know, it's a it's a nine sigma, like, you know, if you are in, in the, in the airline industry, and you're trying to think aircraft, imagine the error of, you know, missing out when did the aircraft need maintenance, right? It has to be like, 100%, close to 100% accurate, they're also not 100%? Because, you know, six sigma is like point 000001. Right? So I will say, there are things where it's, is it life threatening? Is it reputational loss, there is an order. And I think business takes priority as to what they want to risk with. And as I mentioned today, or in the past, a lot of these things were done by hard coded rules, fixed rules, today, data is helping there. And what I see is, it's going ahead with a combination of still fixed rules plus data together, because at some point in time business would say, I'm still learning. So I will say, over years, hopefully data will take and our decision will also be automatic. But depending on the threat of life, or the reputation, or what needs to happen, I think business will decide what to do what not to do. And I think it's an ongoing thing.

Absolutely. And just I mean, this is an interesting point, I think he put up a secondary question saying, you know, businesses are moving targets, they're not you know, stationary things. So, is communication a way with the business to solve this problem of you know, goals may change by the data model may still have not evolved? So, how do you think how do you reconcile these two things?

So, I will say communication with the business is absolutely important. However, one thing we have to keep in mind that I think as I mentioned, the data science is still it's still not reached its peak understanding for everyone. So I think like business folks, there are a lot of business folks who are also learning data science, right, the business owners and you know, business sponsors. So I think today there are two gaps one, we are as I mentioned, somewhere during the journey that business this AI translators are the next generation product managers should be people who know AI. I think people who know AI will translate into good product managers because future product managers can't just cannot be just running scrims if you are running an AI specific work stream, we need to understand AI to the extent where we can explain and talk to the business and we should be meaningfully able to convert that and and kind of eliminate Any such doubts like this should not be solved, this cannot be solved versus saying, Okay, let me I have data, let me go find it. So I feel like that business translator role. And you know, educating our business sponsors, both of them is kind of one of the goals of Chief Data Science officials of Chief analytics officers, I think, and that's an ongoing thing across the industry.

Absolutely. Thank you. We go ahead quickly. There's a question from Martha, we, you know, who asks, How does one switch to a data science or as a career after a few years of experience in the software industry, but not specifically, as a scientist? Most of the times, you know, employers look for relevant experience, you know, what would be a good strategy, according to you? She's currently undergoing the computational data science program with ISC and talentsprint.

Yep. So I was I just, you know, as we asked you the question, I moved back to the slide in Japan of reference to, you know, to move back for reference that mother, we like, I think you should assess yourself, where are you, you know, and it should be a self assessment, maybe working with your mentor or somebody or if somebody is evaluating, where am I currently in which stream. Now, as I mentioned, if you are not very specific about this data scientist, which is here, which I personally believe should not be focused of anybody, everything here is variety of data science slash API slash ml roles where we should try to jump in, because without this, we might have data science algorithms, but they may not be seeing the end of, you know, they might be seeing the end of light or an end of tunnel kind of thing, because a lot of things are just insights. So I feel like, the best way to break into this is, one is completing pretty good courses. The second is, you know, participate in kaggle competitions, you know, participate in open competitions and create a portfolio of data science projects. And I think that should give the confidence for you to jump into data science. And I think the demand is such high that if you are, if you are really pursuing your career, ongoing, ongoing curriculum, and everything I believe anybody could break in, you are still a part of technical industry. I know there are a lot of people may not be a lot, but at least quite a few of the people who have moved from very non technical background to jobs in data science. So I'll say, I am happy that you are you're not just asking the question, you have taken the first step. And I feel like you're on the right track. So just continue. And I think it should work out to kind of find the right job, the only thing is, understand where you should apply. It is not this data scientist, which just says data scientist is a very different meaning out there, which a lot of times so you don't end up in the right place. It's like, hey, I want to be that a lot of them that may not be right, instead of figured out like it might take me a few years to start as a business analyst or maybe a citizen analyst or business intelligence analyst with the skills of data scientists, and I'm willing to be considered early in the industry, not like I have tons of years of experience in something and then, you know, over time, make the transition to a full data scientist. And I think it's a long career and rewarding career sciences, all the efforts you putting in should come to practice, this could be a good, really good map to help us look at it.

So thank you, for them. That's that's actually a wonderful. Go back, like you said, towards people looking at a career over there. We will go on to the next question, I think is again, your reference to the last slide that you had. Kavita asks, you know, she did not get why you prefer quantity over quality, quantity over quality in AI. You she wants you to elaborate a little more?

Sure. I, I remember, it was more on the lighter note, I think a lot of times what happens is it's the quantity here refers to algorithms means data. And quality refers to the software aspect of let's say, we just answered a question. The trade offs, the trade offs is not in the data, the status is not quantitative, the trade offs is understanding qualitative, really, what will make the difference for the company will is is it a reputational risk, or some things I meant to give this to a lighter note, obviously, there are two different disciplines when I go into a very different thing, but here what it meant was, you know, let's not just everything is not quantified. It says here is a not everything is gone, because people are saying who prefers quantitative or qualitative and the name says that everybody prefers but when somebody asked why do you prefer, they actually have to give a qualitative answer? So it seems like both of them go hand in hand, and there is no one over the other.

Thank you. Thank you for that. So there's a question continuing question from Kavita itself. Is it advisable for a mid level career professional to take a you know, take up a data scientist to seek opportunity in the field of AI? This is a very standard question we get across any of these sessions that we do.

And I would say the first thing I'll again as I mentioned, you know, it's it Not one thing is just stepping back. Data Science is not all careers means I will just go and I know it's a data scientist session and everything, but I definitely say career is a very different thing. So you know, evaluate what your skills are, and then go back and take a look at one of those bits. If you just search Gartner, you know, Braden, everything, you will get a great in which industry you are, what is your current professions? What do you do on a regular basis over the last six years, 10 years and seven years, and decide, and as I mentioned, our lens of data science has to change. There is no one type of data scientist and that's the most important thing, I think we are, we are, we are probably at a stage where you feel like that that unicorn concept of data science, which is gone, which I think we are looking at variety of roles within data science, I believe anybody can make a cut who wants to make a cut, but the price we have to pay is tons of education, and tons of time invested. Because the competition is heavy, and the right talent, knowing the right algorithm is important. And not just marking up data, you know, the algorithms, it's like, I will to a genuine advisor, be learn take trainings in a do sessions, you can nowadays get online degrees as well, I think. But however, the ongoing online discounts like talentsprint, and some of those, like tie ups with ISC, it big brands, I feel like they are powerful. The second is continue portfolio projects, I also build a portfolio, I will say, and I think they will be proposing the same thing as well build a portfolio outside of the curriculum. Also, there is a kaggle changed people slide a lot of people who are actually calculator stop calculators, they don't take up jobs, they have so much Excel by just solving kaggle problems that they have. They are actually not consultants, I believe, like, you know, they kind of people are just looking out for them, Hey, can you solve character with this character, I will say, go solve problems, whoever wants to break into AI, data science analytics, it's problem solving at the end of the day. So try to learn more problem solving, not just coding, at the end, a lot of the courses that we do, we try to just pick up the code, it will not help. I think you just we just need to get the essence of what the professor is trying to say what the you know, the consultant, the industry experts, the mentors are trying to save when you are doing that course it's problem solving at the end of the day CEO is looking at those objectives and seeing where can you add this? He's not asking how many codes how many lines of code, have you written? We think the code is tough part and to answer connecting to another question. This is the reason why that auto ml type of things are needed. Like the world is saying, We don't want hardcore coders, citizen data scientist is good, the person is not the coding, the person understands the algorithms, and the person understands the business problem. So he can actually use auto ml to kind of just, you know, get the data will run the algorithm and go share the insights with his business manager. So absolutely, yes, could be my answer.

So again, an interesting question that, you know, probably asked, What are you currently doing at Fidelity Investments? You know, what are some of the challenges that are high level you're trying to solve using data science in your day to day role? Yep.

So, you know, at a broad level, I'll say, right now, my role, as I mentioned, you know, is I lead a bunch of in a talented data science, data scientist, being the, you know, the AI Corp leader. And fidelity, as we know, is in the financial services into, you know, in the health, health, health and welfare domain. So our goal is in a similar to what we see on on general, General financial services, there are risk and mitigation problems we're trying to identify, which What is it fraud, what is at risk, we're trying to identify which customers likely is interested, we're trying to identify, you know, what is how data can be beneficial with regards to what benefits are needed by the company. And as you know, fidelity is a US based company. And I think the focus is the retirement the goal of AI is to help people, you know, get insights and you know, meaningful insights for themselves so that they can decide what's the best for them to retire. So all the all the work that we do is to help our end customers, you know, get meaningful recommendations, which will help them do better in their financial lives. So right now I'm leading a team of data scientists, as I mentioned, we're focused on different facets of the industry and different facets of different types of algorithmic problem that I just talked about in one of the slides like different machine learning. Use Cases is word so you can pick up any of those and you will be working on like if you This is the place in the financial services. So risk analytics and regulation, customer segmentation, Intelligent Automation, great wordiness a lot of those aspects are being addressed.

Absolutely. Thank you so much. I think we are past time. So if can we take just one last question behind chatter. So a lot of the questions are around in terms of you know, you know freshers trying to apply break, trying to get into the first job. etc, you know, and most of the questions usually that we get around, you know, is the job available, etc, etc. What would your advice be to be young lot of people who are there, you know, somebody who's probably just broken in maybe the first second year of his career, or somebody, you know, fresh out of college, what would you advise them as somebody you know, who who leads such a large function for a large organization? What are the two or three key things that a young professional should focus on?

I will say, I will say this is if you are very early, you have an advantage. And the advantage I would say is, and I that's why I came back to this, and I had thought through this, and I thought some of the sports kind of in a relay, and I'll, again, show this, but I'll see what I what I mean by this is, if you're very early on, try to go for education while at work, you know, try to pick up advanced advanced courses on computer science, mathematics, you know, be machine learn today, probably three years, five years back, we never had a course on data science and AI, today there is, so pick any of those good courses from a good brand. And just, I think the path will automatically come however, only caveat I'll call out is it's not a generalized fee. It's if you are very early, you should try to start with a generalized to get a feeling of it. But very soon, you should figure out a specialization you want to do. And the specializations you can think of, I think I kind of showed at a broad level. And it's, it's an interesting way, look at this, these all fields will excite you if you learn more. And I think some of these courses, whatever we have, from top Institute's, they actually address each of them. And there's a meaning in each other's eyes. So my advice is, go sign up for courses, go build a portfolio, believe it's not a one year journey. That's the first thing I'd say it's a multi year journey. And I think that will, that will make the break, I think you will get a good job, it might start with a business analyst data analyst. Don't be heartbroken, everybody starts there, even if they're called data scientists, some places. So I will say the advices, learn learn, learn, practice, practice, practice, and then understand and pick up a domain and pick up a specialization. I think that's my advice to all of you that it's just an growing field. And this field is only going to need more and more data scientists. And the definition of data scientists is somebody who helps make sense of data. Very simple. Rest, everything is complicated. But if we are only focused on that, only way to go up is the first code that I started, you know, skyrocketing everything. So hopefully I answered the question.

Yeah, absolutely. Absolutely. And yeah, I think in line with what usually we tell people that, you know, build the skills up, the job will follow. I think most cases, what we look at is that I'll get the job, but I will not concentrate on the skills. That's not the way the market works. But thank you so much, bihon. It's been lovely having you, you know, very interesting session that we had today. I'm sure we would have loved to have you for a longer time. But then, due to time constraints permit, do not permit us for longer session. We hope to have you back again sometime in the future for another interesting conversation. Thank you, everybody, for joining us today, as well as spending some time on a Friday evening with us. And we look forward to more sessions like these with more experts. Thanks, Ron. Thank you again for spending time. It was lovely having you.

Thank you. Thank you. Thank you. talentsprint. I would definitely like to thank Rami as well enough for the connection and thank you all the participants for the questions. If there was any question that you needed some help or answers drop a note to read through I'll try to answer them back at some point. But you know, happy to have this Friday end on this note. You know glad you all liked it. So let me know if there's any feedback or throw you hear anything so happy to join back at some point in time as well.

Absolutely deserved. Thank you again, all the best. Stay safe everyone. Good night. You

Watch the entire interview here https://www.youtube.com/watch?v=xvlqfxsI1_w