There were 3 rounds out of which 2 were technical and 1 was HR.

Round 1 - Resume based

Introduction.
Little Resume projects description.

I has RAG so I was asked about different types of evaluation parameters like MRR and NDCG, precision and recall and when to use when

Bagging and Boosting techniques, how to reduce overfitting. Bias and variance tradeoff in ML.

Explain backpropagation and softmax function.

Pandas code to find count of null values, SQL code to find emails that appear more than once.

Puzzle - You have 9 balls 8 are identical and 1 is heavier, find minimum times you use weighing scale and find the ball.

Round 2 - DL

I only got round 1 but I asked other 3 people who got round 2, all were asked about PyTorch and other libraries first then a puzzle simillar to round 1.

High concepts of RAG intended towards its system design - prompt caching, multiple prompts to handle same service?

If there is a model with single feature and has 90% accuracy what can you predict and what is the relibability explain everything you can think off.

Round 3 - HR

No idea about this one, only 1 person got this on my campus.