Executive Espresso with Principal Data Scientist, Kranthi Mitra

When you hear the word ‘scientist,’ what’s the picture that pops in your head? Lab coats, test tubes, experiments? A Data Scientist at Swiggy is not very different. Sans the lab coats and test tubes, the Data Science team at Swiggy is constantly pondering in the realm of data and building AI/ML solutions to make the Swiggy ecosystem intelligent. While most of us are reminiscing about a delicious dish we ate recently, Data Scientists at Swiggy are reimagining how an AI system will define the recipe for your next, all-new tasty dish! Read on as Principal Data Scientist at Swiggy, Kranthi Mitra, spills the beans on how our Data Science team is driving most of the predictions and decisions when it comes to dish recommendations, supply-demand estimation and much much more.


As a Data Scientist, how would you describe your experience at Swiggy as against your earlier assignments in other companies?

It has been an exponential journey right from the start. From using data-driven culture to building strong Data Science practices, to driving the AI-first vision. This holistically reflects everything we practice in the space of Data Science. The most profound aspect is the real and immediate impact that the work data science creates on improving customer experience, minimizing cost and much more. Simply put, this is the nonlinear impact that DS has. Uniquely, unlike at most places I worked at in the past, Data Science at Swiggy cuts across disciplines — from usual classification modeling and forecasting, to utilizing ‘optimization algorithms’ and application of computer vision.

We know that Data is at the heart of all that Swiggy does. Can you shed some light on what the team does to build intelligent solutions for consumers, restaurant and delivery partners?

Field visits are essential to get a real pulse of what occurs on-ground. The team frequently visits restaurant kitchens during peak hours, or at times,  take up order deliveries along with more experienced Delivery Partners, or sometimes even visits areas where we suspect fraudulent patterns. All this helps us stay in-tune with reality, and make observations from various modeling efforts. Net-net, we are able to quickly validate if the data is truly reflective of what’s transpiring in the real world. Once the validity of data is established and key focus areas are identified, we jump into solving the problem. This is driven through focused groups, consisting of Data Scientists, analysts, engineers and product managers.

Once the right metrics and objectives are narrowed down it’s about decomposing the problem into its mathematical form. This enables Data Scientists to pin-down on the right modeling problem to solve for. For instance, if the objective is to reduce the number of order cancellations resulting from a restaurant closure, the mathematical formulation will lead to identifying the probability — restaurant being closed — as the appropriate modeling problem. By involving other stakeholders from the very beginning we are able to push the models for production consumption in a seamless manner.

There have been instances where we experienced extreme customer pain like sleeping hungry because an order got cancelled at 2:00 pm, or being embarrassed because non-veg got delivered instead of veg. These extreme outliers, while unfortunate, pose an interesting problem for us in terms of building intelligent systems having the ability to address the most extreme of cases.

Tell us a bit about the Kranthi Mitra that Swiggsters are not too familiar with. What keeps you ticking outside of work?

I love binge watching cartoons with my kids, and YES; even I’ve been given a curfew on that every day [laughs]. I’m also a huge Avengers fan and don’t miss out on a single opportunity to watch them with my kids. I love to learn basic Math and Science like any other kid and find comfort in rediscovering the joy of learning with my children. It’s just an amazing experience and feeling .

The world as we know it, is coming to an end and you are given a choice of stocking up on only two dishes. What would they be, and why?

I’d rather build an algorithm that will choose a specific recipe for me. I’m assuming drinking water is a given, apart from two dishes.

Any message that you would like to share with Data Scientists who are keen on joining the Swiggy family?  

While algorithms are getting more sophisticated and capable of leveraging data at scale, I find that many aspiring Data Scientists lack either fundamental statistics or pure love for data — a natural curiosity to play with it. In my view, these two are highly underrated qualities of a good Data Scientist. If you have this hunger in you, then Swiggy is the all-you-can-eat place for you.

Facebook Comments