I recently had received rejection mail from Google for SWE III (ML). Here is my interview experience:

Rounds:

1. 3 coding round (DSA)
2. 1 ML domain round
3. 1 googlyness

I had first call around 7th APRIL 2024. Asked one month time for preparation, but due to some family issue got escalated to 2 months. [solved 500+ leetcode] zero leetcode questions before.

First round:

1. a medium to hard level question from binary search - successfully coded and discussed the approach, FEEDBACK was very good.
2. a question based on string, due to accent of interviewer, was very difficult understanding the problem. and coded the solution. Since I had to discuss a lot, she gave me Below average feedback. Also she wasn't able to ask the follow up questions so that may be responsible for feedback
3. a standard questions based on design, successfully coded the solution, and answered the follow up questions: Good feedback

got 10 days time for ML preparation:

Prepared ALL classical ML, DL algorithms like CNN, RNN, ANN, read mathematics, derivation, and coded the algorithms from scratch. (not RNN and CNN). Read about all popular loss function, evaluation metrics, overfitting, underfitting etc.

Round 4: asked questions based classifying 500 different types of trees based on images.

Discussed the approach, Loss function, and evaluation metrics.

But interviewer was highly inclined towards making it as product, questions like, how will you enhance USER Experience, How will you decided if the model is good enough, and some other questions like why model is trained faster on GPU than CPU? [although I answered like parallel processing is faster in GPU] but again why?

overall: bad feedback in this round.

So recruiter asked me to reach out after 6 months.

4 months of preparation -> NOTHING (apart from learning)