Google

It is a couple of hard technical rounds. Mostly, you are evaluated on how you think, not if you are able to solve the question. For example, if you brute force and give an answer, it won't work.

The initial round was 45 minutes and included a DSA question. Round 2 focused on DSA and Googliness. There was a third round, but I have not yet received confirmation for it. The questions asked were medium to hard on LeetCode.

Well, MLPs work great in classification problems. The reason we used CONV layers earlier is that MLPs lose a lot of valuable information when trying to extract features from the image, because we have to flatten the image before feeding it to the ne