Shavak Agrawal. Photo: Kumar/Mint
Shavak Agrawal. Photo: Kumar/Mint

Six lessons from a post-millennial on getting a job in AI

Shavak Agrawal says that one should have a good understanding of linear algebra, probability, statistics and other core maths concepts as machine learning involves complex functions with different variables

As a graduate it is not easy to get a job in AI, unless you can show the right experience," says Shavak Agrawal, 21, who works as a data scientist with Microsoft . A bachelor of engineering in computer science from BITS Pilani, Agrawal made sure he opted for machine learning courses during his degree. He had a summer training stint as a research intern at IBM Labs, Bengaluru, and a second one with Quant One Technologies, a Kolkata-based firm that develops trading algorithms. For his final- year project, he worked for six months with Flipkart in Bengaluru, building AI frameworks to identify and predict anomalies in the e-commerce company’s supply chain. Here are six things he learnt:

■ Introspect early about why you want to get into machine learning. This is something interviewers ask and you should have an answer for it. For me, this was triggered by the famous Target store case study, where an algorithm on shopping patterns uncovered that a young woman was pregnant even before her family knew.

■ Keep yourself constantly updated on the latest research. I follow, which publishes the latest research papers, and computer scientists like Yoshua Bengio and Yann leCun, besides blogs like

■ Do online courses.The machine learning courses on Coursera by Andrew Ng, co-founder of Coursera and adjunct professor at Stanford University, is a great start.

■ Spend time on understanding algorithms and how they really work , instead of accepting the algorithm as a black box.

■ Be selective about the kind of job you choose. Today there are AI jobs advertised everywhere, but you need to look carefully and in detail at the kind of work or product that is being worked on and at the background of the people working on it. It’s important to have good mentorship—someone who can teach you on the job.

■ Have a good understanding of linear algebra, probability, statistics and other core maths concepts. Machine learning involves complex functions with different variables. You tweak different parameters and run experiments with them. You should be comfortable doing that.