My experience-

Coding 1

Q1- LC easy and ranked within 50 in top Facebook questions when sorted with frequency. Coded and simulated it very quickly, and talked about the time complexity. Then, I was asked a common follow-up to that question, we discussed the solution. I was not asked to write up the code.

Q2- The given problem was a link list problem and it could be divided into two steps. The first step should be an LC easy- find duplicates in an unsorted linked list. The second part was an LC medium. I did it with two separate methods. After completing the code, made the dry run. Then we talked about space and time complexity. 

Coding2

Q1- LC Medium and ranked within 10 in top Facebook questions when sorted with frequency. Coded and simulated it quickly. Talked about different approaches, and trade-offs. 

Q2- LC Medium and ranked within 25 in top Facebook questions when sorted with frequency. Coded and simulated it quickly. Talked about different approaches, and trade-offs. 

Coding3

Extra round with a shadow interviewer. Performance of this round does not count- that’s what I found in different threads and posts.

Q1- LC Medium and a little bit uncommon level order traversal problem. Could be in top 100 or 150. I got the main idea but coded it to find out I misplaced a logic. I had to dry run the problem twice to see whether my approaches are correct or not. No hints from the interviewer. The interviewer just skipped to the next question.

Q2- Tagged LC easy but never felt it easy. Other people also complained about it. Problem was related to a sorted array. Proposed a linear time solution and coded it upon the green signal from the interviewer. Completed the dry run to show it works. Then the interviewer asked whether any improvement possible. I said a O(LogN) solution might be possible. Then the interviewer finished the interview. If I was asked about it before starting coding, I think I could come up with the O(LogN) solution.

Behavioral

Regular experience. I have given other behavioral interviews too, and I could not feel much difference here.

ML Sys Design

Asked a pretty a simple ML system to design and lead the discussion. I felt it went average. I could not lead the discussion properly at some points. So, the Interviewer had to ask other questions to continue the discussion. I felt that I knew so many things but a few of those did not came to my mind before the interviewer asked about it and I missed some as interviewer did not ask about those. May be, I should have more mock interviews. Also, I could do a better job at feature engineering.

Result-

Rejection came from the candidancy review. It was hard for me- I was expecting a positive result as I have seen other people got the offer with similar performance or at least a follow up. May be coding 3 was counted towards my result. I also think interviewer at coding 3 did not do a good job. It was my bad luck that I did not get some hints from them.