There were 2 in person interview. First was DSA with 2 medium-hard questions one was one graph traversal but confusing with DP 2nd one was on 1-D DP. The ML Breadth and Depth round was on mostly self-attention, transformer, encoder, decoder architecture, deep learning training and some classical ML algorithms.The deep learning concepts of overfit and underfitting conditional question was asked at the end of the interview. I would rate this medium-hard level.