Mathangi Sri works as the Head of Data Science at PhonePe, India’s leading e-commerce payment system and digital wallet company. After completing her undergraduate course in electrical and electronics engineering, she completed her MBA in marketing and finance from the National Institute of Technology, Tiruchirappalli. She started her career in analytics at GE Capital (now Genpact) and later worked in companies such as HSBC, Emirates Bank and [24]7.ai. 

With more than 15 years of experience, Mathangi has a track record of building world-class data science solutions and teams. Some of her areas of expertise include machine learning, NLP, web and text analytics, predictive modeling, business and customer analytics, data mining and more!

Today, she has 20 global patents in data science with 10 global patent grants to her credit. Analytics India Magazine also recognized her as one of the top 10 data scientists in India. Mathangi Sri is an inspiration for all young and aspiring data scientists. We recently interviewed her to learn more about her journey, work, go-to resources and favorite tools! 

Here’s the complete interview. 

How did you start your journey as a Data Scientist?

I started my career in analytics through GE Capital (consumer finance) more than 15 years ago. I had the opportunity to build predictive risk models that actually were making real-time decisions. In addition to this, I also used to analyze and prepare dashboards. I had my foundations in statistical modeling and data handling. I think these are the two strong skills that one needs in their data science career. The distinction between data science and analytics came in the last few years or so. So it kind of became a natural progression for me. The kind of problems we used to solve using SAS and statistics, we are now solving with Python and machine learning.

What are your go-to data science books or resources?

I usually learn about new concepts online through blogs and videos. Some of the blogs I follow are:

  1. Machine Learning Mastery
  2. Data Science Central
  3. Analytics Vidhya

I also follow data science blogs on Medium. For all the latest trends in data science, I follow Analytics India Magazine. I also refer to the videos of Prof Jeff Heaton of the Washington University.

What are the top 3 skills that matter the most to you as a human of data?

  1. You should have the ability to understand that data is at the core of your skills as a data scientist: You have to touch and feel the data before jumping into any algorithmic decisions.
  2. Understanding of business needs is very essential to do elegant problem-solving. Sometimes we tend to make a “mountain” out of a molehill and still not solve the problem at hand if the domain is poorly understood. I believe every data scientist should wear the hat of a consultant first to understand the issue and then go about solving the problem.
  3. Math skills are also important to know the contours of the business problem and the nuances of data you get.

So, in my opinion, the ability to deal with data, understanding of the problem and the skills to solve the mathematical problems should be the core skills of any data scientist.

What are some of the data science problems that you have worked on?

I have built several products/solutions like:

  • A “simple-to-use” text mining engine
  • Data-driven designs to improve user experience in a data-driven way
  • Personalization solutions
  • Text-based “recommendation engines” for customer care executives

What is the coolest tool, library or hack that you recently discovered?

I like neural networks for the simplicity of the algorithm, the flexibility and “transferability”. I think transfer learning pretty much accelerated a lot of applications like image processing.

Describe yourself in an emoji and why?

I tend to listen a lot before I speak up and hence this could be me. 🙂

Author

Love all things data, technology, and startups. Tweets at @leenasoni_.

Write A Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.