It was of the model of 3 + 2 interviews. You will be called on the basis of performance in first 3 rounds.

There was no behavioral, all coding in my case.

Round 1: Fire starts spreading in all 4 directions from a point a, b. You have infinite supply of firemen. They start moving from a point x, y. Grid[i][j] denotes how many people are there in that grid. If fire has reached a coordinate, you can't save people in that coodinate. Return max number of people you can save. This was the round that cost me the most. I required taking hints. The thing is he was not clear with his question explanation. First he told me that there is only one fireman and then he said that No, it was wrong, you have infinite supply of fireman. I was thinking in that old direction for around 20 odd minutes. I required taking hints 2 times but was able to solve and write complete code. It was not that the question was very hard. I felt that he did not explain the question nicely enough and that resulted in me assuming things. I could have also repeated the question to make everything clear to me and check if we are on the same page right at the start. I did not repeat this mistake in the other rounds.

Round 2: You have threads which have list of dependencies. A thread can only be executed only if all it's list of dependencies are executed. What will be the max size of thread pool required to satisfy all the thread dependencies? I was able to solve it and felt really good. It took me time and no follow up was asked as no time was left.

Round 3: Asked 2 easy grid DP questions. Did both. Asked a strings question with only 15 minutes in hand. Given 2 strings x(smaller length)and y(larger length). Return t/f if you can change x to y. Both contain only 'a' and 'b'. Operations to perform:
1) Append 'a' at the end of x
2) Reverse x and append 'b' at the end

I could only tell brute force and suggest that we can also use Rabin Karp to compute the reverse hash. If anyone can solve this optimally then please share.

Feedback: Reason for rejection was

1. Slow coding
2. Requiring multiple hints

I just agreed to what the recruiter was saying but I felt that I was good in speed. And in the first 2 round the questions were a bit of hard level and required around 15 minutes time to only understand the problem statment.

I had a mock interview with one of the Googlers I know, before the onsite. He said that my speed was good and make sure that you are quick with writing code as it was one of the biggest factors in selection. I felt that I was fine with speed during onsite. But there are always things one could have improved on.

General Advice:

1. Just enjoy the process and shift away from focusing mostly on the results :).
2. Google expects you to do atleast 2 questions in almost every interview. I was conveyed this both before and after the interview.

I enjoyed around 4 months of regular LeetCode, which I will still continue. Also really enjoyed giving the interview. The question set was also very good. The feedback provided was a lot better than most companies. I asked if I can get an interview-wise feedback. But I was told that to protect the interviewers, they hide this detail.