Applied online and received a call for an initial phone screening 2-3 weeks later.

Got an invitation for an onsite interview 4 weeks after that. The onsite consisted of back-to-back interviews with 4 engineers, plus 1 HR round.

1st: Non-linear functions in DL: what happens when we apply 'x' in the function? Writing formulas of various loss functions in ML. How to handle imbalanced data. What is KL divergence? What's the benefit of multi-modal? Drawing derivatives of activation functions. Coding: Backtracking problem.

2nd: CNN architecture details, 3-4 coding questions. How do you identify a filter using convolution operation when there is no direct access to filter information? How to deal with audio time signals in deep learning? (Explained pre-processing steps to spectrum).

3rd: TCP/OSI 7 layers, difference between TCP/UDP, 1 coding question. Convert to class instance. Writing a bash script that parses from directories and distributes outcomes in some directories. Explain some Linux commands. Resume questions. Write a program to evaluate ML models' criteria for evaluation: whether to allow this model to pass or not.

4th: Some basic machine learning questions and VGG, Inception, ResNet. Python-specific questions (Generator, Decorator).