In this episode of the Applied Intelligence podcast, I had the pleasure of sitting down with Tanika Gupta, a leading voice in data science and AI. We explored her fascinating career journey, the critical role of data-driven decision-making, and how generative AI is reshaping industries. From fraud prevention to product recommendation engines, Tanika shared powerful insights on bridging the gap between technology and business.
From Curiosity to Data Science Leadership
Tanika’s path to data science wasn’t a straight line. She began as a naturally curious student, deeply invested in learning. Initially uncertain about her career direction, she eventually found her calling in analytics while working at American Express. There, she helped build fraud detection models that saved millions of dollars, proving that AI isn’t just about algorithms—it’s about solving real-world problems.
Her experience spans industry giants like Mastercard and JPMorgan Chase, where she refined her expertise in predictive modeling and AI-driven decision-making. Now, at Sigmoid, she leads teams that develop cutting-edge AI solutions tailored for enterprise challenges.
Generative AI: Beyond the Hype
One of the most compelling parts of our conversation was Tanika’s take on generative AI. While many are still trying to grasp its full potential, she’s already leveraging it to transform business processes. She shared a recent project where her team built a generative AI-powered product recommendation engine for a major B2B chocolate company. By automating SKU recommendations, they reduced the sales cycle from weeks to mere days—saving time and unlocking revenue opportunities.
Bridging the Gap Between Data Science and Business
A key theme in our discussion was how to make data science more accessible to business leaders. Tanika believes that technical teams must go beyond just building models—they need to translate insights into strategic decisions. Her approach? Keeping business leaders involved throughout the AI development process, ensuring that every solution aligns with commercial objectives.
She also emphasized the importance of looking beyond traditional performance metrics. Instead of just optimizing for accuracy, successful AI solutions must drive real business value, whether through cost savings, efficiency improvements, or new revenue streams.
Diversity, Inclusion, and the Future of AI
As a woman in tech, Tanika is passionate about fostering diversity in AI. At Sigmoid, she actively works to remove unconscious biases in hiring and promote equal opportunities, particularly for women re-entering the workforce. She also highlighted the critical role of responsible AI development, ensuring that models remain ethical, unbiased, and aligned with real-world needs.
Key Takeaways from Our Conversation:
- AI must solve real problems. Generative AI and machine learning are only valuable when they drive measurable impact.
- Collaboration is key. Data science isn’t just about algorithms—it’s about working closely with business leaders to create meaningful solutions.
- Bias in AI is real. As AI adoption grows, ensuring fairness, transparency, and inclusivity will be crucial.
Final Thoughts
This conversation with Tanika was an eye-opener on the evolving role of AI in business. Whether you’re a data scientist, a business leader, or just someone curious about AI, there’s a lot to learn from her approach to problem-solving.
You can connect with Tanika on LinkedIn and dive deeper into her work at Sigmoid. And of course, don’t forget to subscribe to the Applied Intelligence podcast for more discussions on innovation, technology, and the future of AI.
Transcript
Imtiaz Ahmed [00:00:02]:
Welcome to Applied Intelligence, a conversation at.
Tanika Gupta [00:00:06]:
The intersection of people, technology and getting stuff done.
Imtiaz Ahmed [00:00:12]:
And now here’s your host, Imtiaz Ahmed.
Tanika Gupta [00:00:18]:
Hi everyone. Welcome to the Applied Intelligence podcast. I’m in Bangalore today and I have the pleasure of being in the Sigmoid office where I’m going to be interviewing Tanika today. Welcome to the show, Tanika, how are you?
Imtiaz Ahmed [00:00:29]:
Thank you, Imtiaz, thank you for having me. And I’m very good, thank you.
Tanika Gupta [00:00:33]:
Super cool. So the first thing that I do on my podcast is I love to ask this question, which is if you had an autobiography of yourself and there were five chapters in that autobiography, what would the chapter titles of each chapter title be?
Imtiaz Ahmed [00:00:49]:
Very interesting question. So if I have to think about. Right, of course, I’ve just spent around some years of my life not going into my age. So I started the first chapter would be the cocoon of curiosity. So I’m a single child. I was born in Badhinda, which is in Punjab and raised in Hyderabad. So being a single child, I was, you know, brought up in a very protective environment and I was very, very focused on studies. And actually my parents used to tell me that please go and play because I was so studious, so.
Imtiaz Ahmed [00:01:22]:
Or that, you know, this whole journey it was more focused which actually kind of laid the foundation from where I am today was more focused on understanding the curiosity, on learning things, discipline. So that was the first, you know, the years of my life on the studies were more focused on building that foundation. So that would be the first chapter of. If I have to describe post that what happened is like when I did my 11th and 12th, the second chapter was more focused would be about the uncertainty period. So which came into the picture. So I was again, you know, like being a teenager, when you are in 11th standard and 12th standard, you’re not sure what you want to do with your career and which generally happens, right. A lot of us are just either going into the medical stream or your engineering stream. I was because I loved maths.
Imtiaz Ahmed [00:02:11]:
I was went into the engineering steam by default.
Tanika Gupta [00:02:13]:
I’m giggling because my parents did the same thing.
Imtiaz Ahmed [00:02:15]:
Yeah. So that is in general what happened. Right. Okay. You like, you don’t like bio, you like math. Go, go for engineering. But eventually, like when I started studying and I was not sure, right. I was looking around, I was not sure whether this is the right path for my career, thinking, what? Oh, maybe I can become an IS officer.
Imtiaz Ahmed [00:02:31]:
Right. So I should have taken biology. Actually that phase of my life was very confusing for me. But Eventually I did went for engineering, but I guess I did not probably, you know, I did spend a lot of time thinking about it and did not study a lot. So that was the second phase post that I did my mba and the third phase, which basically is about serendipity meeting opportunity. And after my MBA introduced me to data science and AI. So I joined the fraud modeling team where I was working and I was lucky enough to join a team where they were focused on the prepaid model. So they were.
Imtiaz Ahmed [00:03:03]:
We had these prepaid cards which are the Bluebird and serve in us. What happens is that a lot of underbanked and unbanked population and it’s very difficult for them to get cards. So this particular card was launched for them which don’t have a good credit history or any credit history. Right. So you can have that, those cards. Now we thought that since it’s a prepaid card, we won’t have a lot of fraud. Right. You generally have frauds on a normal credit card.
Imtiaz Ahmed [00:03:26]:
But we were very surprised to see that the frauds are very smart enough and we saw a lot of frauds and we eventually started looking at those cases, build those rules and then eventually that model. And that’s how my data science journey started. Again, a very good experience on how because this was back in 2013 and then again then it scaled up I went to MasterCard, JPMorgan Chase again more focused on fraud modeling. And in MasterCard I worked on a lot of other areas as well in data science. But then that momentum grew, the fourth chapter being the momentum and right now I think the final chapter, again this is in progress and hopefully would be that I’m very passionate about generative AI, the new field which has come up. I’m working on a lot of use cases and building solutions for clients. And I do see like I’m very, very excited about how is it going to change the whole industry. I’m looking forward because the kind of innovation which is happening every day, it’s very difficult to catch up.
Imtiaz Ahmed [00:04:25]:
Right. Every day there’s something new happening. So I’m looking forward to those next years that is from the professional side or the learning side, I can say on the personal side, again, I’m a long distance runner, have completed 10 half marathons. And so that is again one area of interest. I do love running and that gives me a kind of. It’s something which actually helps me, you can say meditate or release stress. So that is on a personal front, that is something which I do enjoy a lot Apart from like reading motivational books again, the. That is something which I really like and I, I love coffee.
Imtiaz Ahmed [00:05:01]:
So I always tell anyone that very easy to make me happy. Just get me a good cappuccino and you can, you know, make me happy. So it’s. That’s all. Yeah, maybe, you know, that is something more about how my autobiography will look like. And next it’s evolving.
Tanika Gupta [00:05:16]:
I love how you intersected serendipity into specifically moving from your MBA into Amex and then falling in love with data science. Mainly because I also fell into data science and looking at all of these technological things because I come from a sales background, I don’t come from a pure data science background, but data connects all of this stuff and then this new field that’s coming out, which is generative AI, I think the most value gets unlocked when you connect the dots between various experiences, various fields to create that value. We’re going to dive much deeper into that. But the first question I had for you is, can you share a moment in your career when a data science solution you developed led to transformational change?
Imtiaz Ahmed [00:06:08]:
Yeah, So I think there are many. So luckily I’ve been able to work on a lot of different solutions which have created an impact. But the one I spoke about, the one which I started, because when we started, the fraud rate was really very high and to work on that solution, the one we built for the prepaid cards that actually led to savings on millions of dollars, that was one of the most, or one of the most transformational data science solutions which I would have built. So the one on the prepaid cards.
Tanika Gupta [00:06:36]:
Sure. What was the insight in terms of. Yes, fraud was happening, but how did you kind of identify or what were the signals that were being processed to find these people out?
Imtiaz Ahmed [00:06:47]:
Yeah, so there are some very interesting things we observed. Right. So. And which you would feel that are very obvious. Right. When you are looking at. So one example I’ll give you. So, so in one scenario, like we were looking at a use case and we saw that what, there’s an ATM card, right.
Imtiaz Ahmed [00:07:02]:
I mean, you could withdraw money using ATM using that card. And we saw that when we started analyzing one case of a fraud, that a customer reported a fraud, Right. They said that they lost their card and then someone withdrew the money from the atm. And then we said that how did you they get the ATM pin? So they said that we wrote it at the back of our card. Right. So. And then again, the same thing happened with that customer. Right.
Imtiaz Ahmed [00:07:27]:
So it’s very obvious, like you’re still allowing that thing to happen. So there were two scenarios where the same thing happened. So this was one case, so then we kind of flagged it out. But another interesting case which happened was that what used to happen as a process that people used to. The fraudster used to get access to the card and then they used to actually. So as a process, right, what we used to do is that we used to send back that the new details or the address details. So basically, if you are rechanging your PIN to the new phone number, not to the original one. So basically the fraudster used to get access to the card.
Imtiaz Ahmed [00:08:08]:
Right. And so basically I’m a fraudster. I’ll say that, you know, I lost my card or I get get access to your card, then I call back, then bas say that I want to change my pin. Now the pin, instead of being sent to the original phone number, they used to update that and send it to the new phone number. So this was a process gap, right? So basically then it’s very easy for the fraudsters to. They figured it out, got access to the card, and they were updating it and then got hold. So one thing I realized, right, A lot of solutions which get solved are solved. When you go deep into the data, you start looking at those cases in details as compared to just building models.
Imtiaz Ahmed [00:08:44]:
So a lot of times I tell my team and a lot of peers as well, just don’t take up data and think about a lot of focus is on different kind of modeling methodologies as compared to looking at data and understanding those trends. So we just take up data and fit it into those fancy models, either generative AI or lstm. Deep learning, we want to use deep learning, but looking at those trends or looking deep into the data is actually which brings value in solving problems. Even today, we have done a lot of advancement in technology, but that is what actually brings value.
Tanika Gupta [00:09:17]:
So what I’m hearing is human behavior and analyzing the human behavior and then using the machine learning and using technology to expose all of those cases where that similar pattern of behavior has happened is a very efficient way. But right now it still feels like humans are needed to kind of find these cases.
Imtiaz Ahmed [00:09:35]:
Yes, yes, I do believe right. So specifically, we do need some analysis. Those are useful. We do have models which help us bring out, help us ease the process. They have generative AI probably could help you, but it’s mostly on text data. But still it has helped you automate a certain process. But human intervention is required to bring out that value. And specifically in those complex scenarios.
Tanika Gupta [00:09:58]:
So in order to come to these outcomes, you need collaboration. So how has the collaboration between data scientists and business leaders come together to solve problems like this? So how do you kind of bridge the gap between the technical knowledge that you need and the commercial outcomes that you’re trying to drive?
Imtiaz Ahmed [00:10:16]:
I think one of the major thing, right to understand, since now I work with more closely in sigmoid, I work more closely with business leaders than in my previous roles. One of the things I’ve learned is that there has to be an empathy, right? You have to understand their pain point. So whenever you are solving. So I used to and that has been a learning for me as well. My focus previously used to be just on, you know, making sure that KPI, the performance metric, right, either the accuracy, those are met. But now working with business leaders, right, you realize that you have to understand what exactly is the business value they’re trying to drive, right? You have those discussions, we are solving a problem, but what is the major benefit or you know, the major value you are going to get out of that. And you design your solution in a way that you are able to achieve that it should. So just not focusing on those KPIs or metrics, but looking at the overall picture or the overall benefit which is going, that is very important.
Imtiaz Ahmed [00:11:06]:
So that is one major thing. When you’re working with business leaders looking at the bigger picture, what value is going to come out, what value they are looking for, what is the major pain point you are going to solve for them? Another thing is having them involved in all the discussions. You are building a model, but having them discuss what are your thought process. You can have those take them through the journey in a very simplistic way, right? Means you don’t have to make it very complex we are using, but you can give them the thought process. What you are doing, trying to solve in a very simple way, keeping them involved in those discussions, keeping them those meetings or those checkpoints to make sure that they are involved in the process and they feel not just giving them in the end. The model which has delivered ensures that you are able to move in the right direction and bring out that value for them as well. So I feel that is the major point. Having them in the loop, looking at the broader picture, keeping them aligned is a way to actually work together and bring value for both of the parties.
Tanika Gupta [00:12:06]:
How do you simplify these things though? Because the math is really hard. But ultimately people are hunting for growth or people are hunting for savings to create that value. What’s your process for like simplifying things down so that a commercial person who’s completely non technical or not has not done a PhD in math can understand the concepts that you’re bringing to life?
Imtiaz Ahmed [00:12:28]:
Yeah, When I discuss these with business leaders, right. Of course they don’t, we don’t go into the technical like this is the model, right? But what exactly it is doing? So if I’m building a model for them, for example for demand forecasting, right. So the concept will tell them that okay, there’s a time series model you are using, right? But what exactly you are analyzing the patterns, historical patterns. So we give them, we take them for an example. Like this is like what has happened in the past, right? And if you look at the data, right, this is what how your trends or you know, how your demand was for this particular product in a particular region. Historically this has been the spikes. These are some macroeconomic factors which impacted and that led to this particular. And this is what we are trying to capture through the model and this is how we’ll actually be able to predict in the future which happen.
Imtiaz Ahmed [00:13:12]:
So taking them to a very simple example and telling them that how model would be able to capture those scenarios using all the data we give it to and then how it will actually be able to generate. So basically it’s a very, very simple process of just taking a few things and taking them through on their data, right? Like how exactly we are going to do is something I generally do to make it easier.
Tanika Gupta [00:13:33]:
One of the funny use cases for me of using ChatGPT is asking it to come up with analogies for explaining multi armed bandit model, for example or other like really hard technical concepts. So like causal attribution for example, because it really understands the mathematical concept and it really understands how to simplify this stuff down as well and then gives you really funny stories that you can essentially tell off the back of that as well.
Imtiaz Ahmed [00:14:02]:
That’s a good point. I think ChatGPT is making it all easier for us, right. And yes, I also use it sometimes to make it easier for me to express my thoughts.
Tanika Gupta [00:14:12]:
So how do you ensure diversity and inclusivity when it comes to your data science teams? Mainly because you know, when you’re working with AI models there tends to be a lot of bias. Right. So how do you account for that and deal with that within your role?
Imtiaz Ahmed [00:14:27]:
Right. So one is the diversity in the team, right. More about, what I understand is more about the hiring and how we ensure that. So as a women leader in my company, right, I do make sure that we give equal opportunities, right. When we have a panel for hiring, we do ensure there are no unconscious bias happening in the panelists who are being selected. And you know, and even when we screen those profiles, we try to actually give more in case there are opportunities. Someone who is returning from after a break. We do don’t disqualify them because they have taken a break.
Imtiaz Ahmed [00:14:56]:
Because that’s what generally happens for women as well. Right. A lot of them take break for personal reasons. So it’s not that they have taken a break so they should be excluded. We, we do give them opportunities to be onboarded. And also one thing we do right is to make sure that from a hiring perspective, the facilities we have in the organizations are something conducive. So we have CAP facilities for women who are traveling because we do have sometime calls which go beyond. We do have a call late in the night and there is a requirement to travel.
Imtiaz Ahmed [00:15:27]:
We have CAP facilities to make sure that they can travel back safely. We have a lot of facilities in house as well which are conducive to make them comfortable in while working in Sigmoid. So we want to make sure in Sigmoid everyone is, you know, there’s a proper working culture and collaboration and there are no unconscious bias for any women who is working here. So that is how I’m trying to ensure that we have that diversity. And now coming to your next part on models, right. Which have. And there are a lot of use cases, we actually in Sigmoid have developed a LLM validation tool which basically checks each response for your ethical bias. Toxicity, Right.
Imtiaz Ahmed [00:16:12]:
Relevance. So it’s basically a validation tool. So this was built for a question answering system, but again it can be customized to any kind of a solution. So every response is tested. So you get a score that how relevant it is to the question. So if you ask a question, right. It uses a RAG framework on a PDF documents to give you those answers. Now, once the answer is received, right.
Imtiaz Ahmed [00:16:33]:
So there is a scored metric you also get. So you know that if there are certain bias, right. So if there is a toxicity PI information is being disclosed or so those are from the responsible AI. And also relevance that how relevant and you know how faithfulness and groundness, which are more from the performance metric perspective. Yeah. So for generative AI I think it is very, very important because this field there is a lot of hallucination which is there. Right. There is a lot of inherent or unconscious bias which can come in.
Imtiaz Ahmed [00:17:02]:
So I think having some kind of A validation is necessary when we are building models. And I feel that is something is very, very important. Since I am like, you know, like I said, I’m very passionate about this field. I do feel that it’s expanding rapidly. There are a lot of use cases and applications coming. But as anyone who is working in this field or building applications, we have to ensure the solutions we build. Right. Because they are the solutions for the future.
Imtiaz Ahmed [00:17:29]:
We do ensure that how we can make them more responsible when we talk about all these from the ethics and privacy perspective.
Tanika Gupta [00:17:37]:
But essentially, should machines monitor machines as in, should AI monitor its own homework? When should humans get involved and be in the loop when it comes to validating outputs?
Imtiaz Ahmed [00:17:49]:
Yeah, so I think we do have like, you know. Yes, you’re right. A lot of these validations are model validating model. And this is because of the way this is all structured. But I do feel human in the loop is important. Right. It may be depending on, again, depending on the scenario we have. Right.
Imtiaz Ahmed [00:18:09]:
The kind of solution we have built and how critical it is or how much while testing we can test out. Right. There could be certain use cases where you do see a lot of hallucination happening. More human in the loop can be incorporated as compared to somewhere which are less critical. And you see it’s performing, the model is performing well. But at this stage right now, I do feel that human in the loop is important and we should have human in the loop not only for this, but also to give that feedback back to those systems so that they continuously improve. So we should have that to make sure that we are building. So since we have not reached the stage, I feel that we can leave it without on just on reliant on machine.
Imtiaz Ahmed [00:18:46]:
So human in the loop is important and again that involvement can be decided based on the, you know, the solution and the criticality.
Tanika Gupta [00:18:54]:
Super cool. So moving on to like process, what frameworks or methodologies do you recommend to ensure a product centric approach to solving problems like demand forecasting or fraud or determining fraud risk, for example?
Imtiaz Ahmed [00:19:09]:
Yeah, in general, right. Like now we are working. I’m working in a service industry. We do work. There are a lot of solutions we built for our clients as a company. Right. We always have this mindset of building like a product and not like in terms of. Right.
Imtiaz Ahmed [00:19:22]:
That we are building a solution just for a. So and what are the things which we keep in mind right. While we are building solutions? First one is that when you’re building that solution, there is a continuous iterative thought process in the terms that it should not be that we are just building it one time. We always keep in mind that there would be feedback received and how we can actually we build in a way that it can be incorporated and we can continuously enhance and improve that particular solution. And also important is that when we’re looking at those, like I spoke about those KPIs, it should not be just meeting that particular KPI, right? We should be focused on around how we can actually drive business values and how we can look at the overall picture. And what we as a company have done is that we have built a lot of accelerators because what we have seen, right, when you build from a product mindset, right? One is that how we can use those some of the codes or, you know, some of the reusable components. So somehow when we are building a solution, there is one part is that there are some reusable components which can be actually be reused across when we build that product for. Across different companies, right.
Imtiaz Ahmed [00:20:29]:
Or different clients. Now the second part is your. There’s some kind of a customization which needs to be done, right? So even that is actually one of the biggest benefit of building accelerators over products in general, what happens is that when you’re building products, right? So it’s more about that you have a product which is built and then you need to work along for different clients. Now what we with the mindset, since working in a service industry, we kind of have those frameworks, as in we build those accelerators in a way that there are certain reusable components. So that actually reduces our build time from 100%, you know, building from scratch to actually 30, 40%. And then we can use those components which can be customized for clients. So we do keep that process in mind when we are building.
Tanika Gupta [00:21:12]:
So services projects often prioritize customization. How do you balance the need for bespoke solutions, fully customized solutions, with the approach of, you know, from a future thinking view of developing accelerators. So how do you kind of balance the need for a bespoke solution for a use case of one into thinking about that solution being possibly used for other use cases later on, right.
Imtiaz Ahmed [00:21:37]:
Whenever we are building those solutions, right? Like I said that some solutions are something like if they are already, we have worked on it, we already are using some of the components. We do make sure that we are not just building for that particular. It’s built in a way that there are certain reusable components which can be actually be used for different projects. And also there is a specific customization. So we do try to take up those learnings. So whatever learnings or customizations are there, right. We actually use those learnings when we are building for other clients. Maybe there are some good things we learned.
Imtiaz Ahmed [00:22:07]:
There could be very specific customizations required for that particular project or. But we can use those when we are actually building it out for them. That is how we try to balance it out. So even when you are building a very specific requirement for a client that could actually become a learning or you know, something which can be used for different projects we might have done in the past. So we actually take back sometimes those learning and go back to some different clients and tell them that there are some new learnings and how we can actually enhance or improve those solutions. So and we kind of build. Enhance our accelerator further as well. So it’s more about every project or any is a learning.
Imtiaz Ahmed [00:22:41]:
You understand certain concepts. Right. And those can be reused and actually be helpful for other projects as well. So keeping that mindset in mind that it’s not just one. And that’s the best part. Right. Because we get to work with different, different clients and different projects. There are a lot of learnings across and that we actually having that learnings and understanding on different kind of ways and different thought process because it’s not just the solution.
Imtiaz Ahmed [00:23:04]:
There are different thoughts when you work with different leaders. Right. Different ways of thinking and from a business perspective and how those solutions could be modified and we actually are able to malate all of them whenever we are working on a new problem or even use them to go back to our existing clients and enhance those solutions further.
Tanika Gupta [00:23:20]:
How do you keep up with all this stuff? Like there’s like a new thing being launched every day, new model, new way of, you know, processing data. Like what’s your process of personally educating yourself?
Imtiaz Ahmed [00:23:30]:
Yeah, so like I said, I. I’m. I’m trying to catch up. I have not even able to catch up. Right. It’s a, it’s a. It’s this journey is I again I agree there’s so much to learn and it becomes quite daunting. Right.
Imtiaz Ahmed [00:23:42]:
Because you feel you’re always lagging behind because every day there is a new model or some new advancement and I do keep. So there are certain, you know, websites and some information sources. So Andrew Ng is one I really like. You know, kind of love him so his deep learning Org he has a newsletters and he has some courses as well. I do follow a lot of these people like Jeffrey Hinton and Andrew Angie on their link on LinkedIn as well as on Twitter to get those latest updates. Apart from that, I do interact with my peers as well. There are a lot of these conferences which I do and attend to understand what is happening, you know, in. In this space and interact with other folks in different who are working in different industries to get those insights.
Tanika Gupta [00:24:22]:
In terms of thinking about generative AI use cases, what’s your favorite use case right now?
Imtiaz Ahmed [00:24:28]:
Interesting. So there are a lot. Well, I’m just thinking which one would be the favorite one. One of the projects we did recently for one of the clients which we are kind of, we are working on is so they had this problem. So we actually there was a hackathon which happened where we had these. Which we competed with other vendors and we won that hackathon. And the problem statement they had was that they wanted to build a product recommendation engine. So this company is a chocolate company, chocolate and coca company and what they do is that.
Imtiaz Ahmed [00:24:58]:
So it’s a B2B company. So they work with all these your different confectionaries, different even brands like Modelase and all. There’s a specific requirement which comes to them from the. The sales will get a requirement on a certain SKU and they have 20,000 sqs and it becomes very, very difficult to figure out what is the right SKU to recommend they go back to R and D which takes again six to seven days. Right. Because R and D has their own process and other priorities to get that recommendation back. Now what we are doing for them is that we are building a recommendation engine which basically is about understanding those needs. So when a customer request comes, we actually create a customer pitch or basically that we summarize that understand those requirements then go back to their data, right and actually try to see that from the basically they have again a huge data.
Imtiaz Ahmed [00:25:46]:
Again we are trying to create a knowledge engine and then use generative AI on top of it based on the customer required extract the relevant information and map it and get the right sku. So that is something we are building on. And the response so we did build a prototype in the hackathon and the response has been immersed because this is actually solving a very big problem for them. So what is happening is that today the sales like I said, there is a huge process of the street to market is less. We also create a sales pitch. So what happens is that once that recommendation comes we also doing cross sell and upsell. And also apart from that there is a sales pitch which gets generated. So for the sales agent, right they also get what are the key Features or the technical details which they can actually use to pitch it to their clients.
Imtiaz Ahmed [00:26:30]:
So this whole sales cycle and there are more things which are planned for the future. But that is the one of my favorite use case because this happened recently and kind of values it will generate for the sales folks because right now they miss out a lot of opportunities. What happens is that if you don’t get the right recommendation, if they don’t get it, put that request for an NPD which is a new product development. So there is a lot of. But they already have a 20,000 SKUs as a company. They don’t want to invest into new product development. They want that how we can actually leverage that existing products which are there in our. With us sitting down in our inventory.
Imtiaz Ahmed [00:27:07]:
Like we have so many products, why don’t we actually kind of recommend there might be a slight mismatch in the customer requirement. Right. So but how we can actually suggest them which is the best closest match. They might be flexible on certain aspect. Okay, my fat percentage is a little less. Right. But I’m fine with this product. It doesn’t match the exact requirement, but it will work out.
Tanika Gupta [00:27:26]:
What I really like about this example is that there are so many steps that you’re cutting out in terms of the human process and for the end user, the sales agent, you’re essentially handing a response to them on the plate so that they can action against that request very quickly. Whereas previously it sounded like they were taking days if not weeks to respond to a customer request, which is potentially the slower you respond, the more likely you are to lose a deal. So that’s very cool. What advice would you give to organizations looking to future proof their AI and machine learning strategies? Because there’s just so much happening. How do you lay the right foundations to set yourself up for success with all of these new things coming all the time?
Imtiaz Ahmed [00:28:08]:
I think first thing is to understand where do you stand, right. So a lot of organizations are in different phase of their journey. So it’s very, very important that you do that assessment, right. To understand that what exactly where you are. Right. And there are a lot of parameters to look at. And we actually at Sigmoid have a generative AI assessment report. So you can actually go there and there are a set of questions we ask to actually help you understand that where exactly do you stand and which areas do you need help in.
Imtiaz Ahmed [00:28:35]:
So I would say that you as a, as a. For any organization it’s very, very important to look at different aspects, right. Do you have the right infrastructure Right. Do you have the right people? Do you have the right training? Right. So there are so many aspects. So it’s very, very important that we look at all these aspects, right. So we. And what kind of problems you’re trying to solve.
Imtiaz Ahmed [00:28:56]:
So it’s very important to think about it because what has happened is I’ve seen that in my interactions with a lot of folks that because generative AI is there, right? Like everyone wants to do it without thinking that what exactly we are trying to achieve. So it’s just because it has become the cool and magic word, we want to fit it into everything. Of course it’s very powerful. But having that thought process, like defining that clear objective of what we are trying to achieve and then starting on it could be that you might want to set up that old team in house. Then it becomes a very different problem as compared to you want to engage with certain partner who wants to do it for you. Now, depending on the kind of scale and the kind of problem you are trying to solve, every problem is different. But it’s very, very important that we have a solid infrastructure in place, right? And to that because again privacy, ethical, you know, those, those infrastructure making sure that whatever we are, the models we are building or the applications we are building are very, very robust. So that is critical factor.
Imtiaz Ahmed [00:29:53]:
And then of course the talent, right? The people who are leading it, the driving it, the building that in house talent or partnering up with someone. So those things are very important because those are the ones who would actually key to drive your solution to success.
Tanika Gupta [00:30:07]:
If you could solve one major challenge with AI today, what would it be and why?
Imtiaz Ahmed [00:30:12]:
I think we kind of touched upon it, like I said about the ethical and how do we make sure that. So hallucination or the ethical and bias, those are some of the areas which I feel are very, very important. We are improving over time. But still, you know, that whenever we are building applications, then like I said, there’s a human in the loop. There is still that some area, right? We are not sure or that there could be a hallucination and the responses could, you know, we won’t get the responses we look for. And we are trying to, you know, improve it through a lot of process. But I still feel there’s a gap. So I just do want to make sure that how we can make it more foolproof, that would be an area I would love to and I’m.
Imtiaz Ahmed [00:30:54]:
I’m sure that would happen with the advancements. And like you said, right. Every day we are having those powerful models now who can do reasoning as well. So we are moving towards that direction and the kind of models, you know, the responses you get, you used to get like one year back and now are so different. Right. So but yes, that would be something I would definitely want because then you know, just imagine like you are very sure like hundred. Although I again epitome of reaching the. Yeah you know that phase in the, in our journey.
Imtiaz Ahmed [00:31:28]:
But then it would like it would be totally transformational. I would wait to. For that day to come. Hoping that it comes soon.
Tanika Gupta [00:31:35]:
The singularity is not far away. So let’s wait and see. So you’re an accomplished long distance runner. Do you find any parallels between the discipline required in running and problem solving using data science?
Imtiaz Ahmed [00:31:48]:
Yeah, I think long distance running, right is very different. And when I did my first run I still remember it was, I actually, I just, it was, it was very difficult, it was not easy and I wanted to stop right Like. But the only thing in my mind was that I have to continue, right. I have to. And it actually, you know, walking became very difficult at that point of time in the end then running because it was first and so you. I had to kind of you know, run to make sure that I complete the eight point. But I think one thing which I learned is right, you don’t sprint, right. When you’re doing a long distance you don’t sprint.
Imtiaz Ahmed [00:32:24]:
You Pisces. Yeah. You have to survive those 21 kilometers, right. It’s not just 100 meter sprint, it’s a long journey. And so having that patience, right and having to make sure that it takes time but you have a goal in mind. So even for data science problem you can’t just solve them in like you know, I don’t see like I told you about that thing about you know just fitting that model and getting that answer doesn’t work. You have to look at data. There is a step, you have to be patient, you have to keep on trying iterating, thinking about new ways, no new features which can be added to get to your correct solution.
Imtiaz Ahmed [00:33:01]:
It’s very important. So that I feel is the, you know, a very something which I’ve also learned to become more patient and having that clarity on the goal on the mind. And yes, I think again, you know, you start enjoying the journey, right. So even with the like long distance, the high you get once you cross that finish line, right. You forget all those pain or you know, the. Of course it’s not easy. Whenever I start I always think why am I doing this to Myself, am I. So it’s.
Imtiaz Ahmed [00:33:29]:
But the. And it’s always the high I get after I complete the run. And there’s the term like runner’s high. And you know, it’s actual. It’s there. You get that high once you cross that line. And I think that same high you get when you’re building those models. Right.
Imtiaz Ahmed [00:33:44]:
So it’s a process.
Tanika Gupta [00:33:45]:
You build and it works.
Imtiaz Ahmed [00:33:48]:
Yeah. When it works and you see that and you get that. Oh, yes, I did it. Like, you know, that is. I think, yeah. Some of the panels I can draw from both.
Tanika Gupta [00:33:57]:
What advice would you give to aspiring data scientists, specifically women looking to make their mark in a very male dominated.
Imtiaz Ahmed [00:34:04]:
Yeah, I think first is that I personally don’t. Although, you know, I feel that everyone is like the same opportunity. Although I do agree that we have less representation and I would really, really want that we have more women. So I think we should, you know, that first is to have that confidence, right. That everything is not about male. You know, I don’t believe that, that it’s, you know, it’s, it’s just for male or, you know, that they only can do it. It’s their field. I believe it’s an equal field.
Imtiaz Ahmed [00:34:31]:
You know, you need to have talent. And even for talent, I believe, right. You just need to put that effort. It’s. It’s a field. If you put that effort, you put. Have that passion, you can learn and you can, you know, grow in that field. So don’t have that inherent bias in yourself that it is a male dominated.
Imtiaz Ahmed [00:34:45]:
And you won’t get that. If you work hard, you put that effort, you also have an equal chance to rise and grow in this field. And again, like. And it’s all on yourself, right? If you believe that you can do it, do it continuously, learn, improve yourself. And one thing I believe, right, like, women kind of this is inherent. Like. And I read Lean in as well. That also helped in my personally.
Imtiaz Ahmed [00:35:07]:
So this was gifted to me by one of my leaders, women leaders, when I was young in my career. And that book actually has a very transformational impact on the way I thought. So one of the things they mentioned, right. There’s a very good example which is there and I would like to share with. So Cheryl Sandberg, who is the author of that book, she shares an example that there was an exam she and her brother gave and she just felt I’ve not done well. And her brother was like, oh my God, I just stalked that exam. Right. I’ve done amazing.
Imtiaz Ahmed [00:35:38]:
And both of them Got the equal percentage after the exam, right. They had the equal marks. So. And that is what in general I have seen as well, right. Women tend to underestimate themselves, right. That you know, even they will do the same thing. They will feel that we are not up to the mark and the men are like, oh my God. And that is what happens, right.
Imtiaz Ahmed [00:35:55]:
Even in your day to day interaction. I feel in my team as well, I have women, you know, who are there and then, but the men will come and say, we want this or you know, we are confident, give us this opportunity. And women are always like not going for it. So I believe that you have that confidence. Even if you’re 50% ready, right. Go for it and ask for those opportunities because that is what men do. Right. So that is an inherent difference between women and I think it’s kind of, you know, the way things, you know, you’re built or, you know, but get those opportunities, get that seat on the table, don’t shy away and continue to learn and grow.
Imtiaz Ahmed [00:36:26]:
And you, you will rise.
Tanika Gupta [00:36:28]:
Super cool. I loved the fact that you shared that somebody gifted your book. One of my questions that I like to close out the podcast with is what book do you like to gift most to your friends and family?
Imtiaz Ahmed [00:36:40]:
So actually there are many I just thought coming to my mind right now. But there was one about this guy who is an Indian guy who started his own startup. And the best thing about what he talks about in the book is right again, creating that value. So it’s about do something which basically you should not sell your time. You should create that either. It is like, you know, the way you have created a postcard, right. Like basically creating something. It could be a code or it could be something which generates value.
Imtiaz Ahmed [00:37:07]:
So if you are selling your time, then you will always be kind of, you know, lagging. You will never be able to build wealth or grow. You would always be at the.
Tanika Gupta [00:37:17]:
That depends on how much you charge for your time. Okay, what productivity hacks do you use to make your life easier?
Imtiaz Ahmed [00:37:26]:
So one, I, I have a time block. So every morning there is a specific time which I blocked again to keep up with the ever growing and ever changing field of generative AI. And I also use Pomodoro technique. So that’s like 25 minutes. That helps it just keeping so in that, that those are. And another thing is that I also create those focused. No, no disturbance, right. That, that in the morning, in some, some time blocks also whenever I’m trying to work, right.
Imtiaz Ahmed [00:37:56]:
I just try to make sure that all the distractions are away and Pomodoro is one of the good ways to do that.
Tanika Gupta [00:38:04]:
Yeah, false deadlines to yourself which is basically the way to get a lot of work done. I personally find like if I don’t have a timeline or a due date I can’t get anything done. So using Formodora is a great way of doing that. Tanika, thank you so much for being on the podcast. This was a lovely conversation. I’m sure everyone’s going to find a lot of value from it. If people want to reach out to you, how should they do that?
Imtiaz Ahmed [00:38:27]:
Yeah, so I’m Quite active on LinkedIn so you can search for Tanika Gupta and you should be able to find me and yeah and I can share my email id as well which is rohini tanika gmail.com cool.
Tanika Gupta [00:38:44]:
All right. Thank you so much for being on Applied Intelligence and looking forward to chatting again soon.
Imtiaz Ahmed [00:38:50]:
Thank you.
Tanika Gupta [00:38:51]:
Thank you.
Listen to the full episode here.
My Social
Watch the full Podcast here
Listen to the full Podcast: