1. Phone Screen: Interviewer asked to open leetcode on my browser, share my screen and solve random question [You cant pick a problem you already solved, haha]. Quite unexpected experience. I got a hard string problem. Managed to get solution that passed most cases and TLE'd  on the last few. Interviewer was happy. For ML part,  Interviewer followed up with general ML and NLP questions
2. Virtual Onsite: 4 rounds
	a. Coding  + ML theory: 1 coding question on DFS + matrix. ML, DL & NLP theory questions
	b. Coding + ML theory: 1 coding question (standard leetcode hard). ML, DL and NLP theory questions
	c. Coding + ML theory: 1 coding question (binary search variation). ML, DL and NLP theory questions
	d. Manager round: General ML/DL questions which test broad understanding of ML and DL rather than specifics

1. Phone Screen 1 : 1 leetcode style problem. Sliding window types. 
	I did not get the correct intuition first. Interviewer gave hint. I used it well and came up with a good solution. Clean code and worked through a few examples as well. 
	Should've barely scraped through I guess. 
2. Phone Screen 2 :1 leetcode style problem. Heap + sorting. 
	Did very well. Got optimal solution immediately. Wrote Clean code and worked through couple of examples (noticed and fixed a minor bug while doing this, interviewer appreciated that)
3. Virtual Onsite: 4 rounds + behavioral
	a. Coding: 1 leetcode style problem: started with leetcode easy. I coded it up very quickly paying attention to naming, indentation, and dry run with example. Then he modified the problem to leetcode medium. Discussed two approaches  (linked lists/hash maps), pros and cons.  Coded linked lists approach. Was 95% done with code. 
	b. Coding: 1 leetcode style problem. BFS with added logic. Got super confused at first. Interviewer helped to get on track. Got the final logic in an 'aha' moment about 15-20 mins into the interview. Feared not having enough time but, managed to code it up quite fast and quite well. Missed a couple of corner cases though. Interviewer pointed them out. I guess interviewer must have given not bad/decent feedback at best. This was my worst round. 
	c. Behavioral Round: Fun. Prepped all standard behavior questions. half of the questions were completely new, half were expected. Did very well.
	d. ML theory : A broad general ML prompt based on recommendation (more like ML system design question). Discussion for 40 mins on various ideas/ pros/ cons. Focused on DL more than ML
	e. ML theory :   A broad general ML prompt based on NLP
	Discussion for 40 mins. Focused on all the latest NLP approaches (transformer, lstm) discussed pros/ cons of each.