Interview Type: Internship

Result: Selected

Difficulty: Medium

College: IIT BHU

Interview Date: Feb 2026

Number of Rounds: 2

Topics Asked: ML

It was around 45-50 mins. Interview was mostly around ML and whatever I had written in my resume. They were asking a lot of "why" questions instead of just definitions. ML: what is token? why tokenization is important? one of my projects was on structured/tabular data prediction, they asked why I used R². Then compared it with MAE and MSE. Like what does R² tell which MAE/MSE doesn't and vice versa. How do you monitor model performance after deployment? How do you handle imbalanced data? Mention different approaches and when to use them.