Google L4 SWE ML Interview experience

Round 1:
Two standard LC Medium tree questions. Solved first one in 20 minutes. Second was little harder. I solved it and almost finished coding.

Round 2:
Strange coding round. Only 1 medium tree question. Finished coding in 20 minutes or so. Then interviewer told me he has no more questions. He asked me to write unit test cases. I wrote some test cases and then we discussed different edge cases for the rest of time. I expected one more question. Not sure why it was not asked.

Round 3:
Machine Learning round. Lots of questions related to feature engineering, loss function, regularization, performance evaluation. Anyone with some exposure to ML can answer them. But few questions I could not answer which were like, write the formula for hinge loss and derive the cross entropy loss function from definition of entropy. I told that I do no remember. So Mixed feelings about this round.

Round 4:
Deep learning round. I was asked questions around CNN, RNN, model selection, hyperparameter tuning, transfer learning, input data preprocessing. Again, anyone who has done few deep learning projects would know them. I believe I did good. But hard to say.

Round 5:
Behavioral round. It went fine.

What are my chances for passing HC?

YOE: 1.8 years
TC: 170k

#google #machinelearning #machinelearningengineer #softwareengineer