Bloomberg was the worst on-site interview experience I'd had to date. I had one phone interview before they called me on-site.

After reaching the office, one of the recruiters gave me a building tour. Following that tour, interviewers took me to a room for the interviews.

1st Interview: 2 on 1 (Technical) This is how it started. The first question was pretty easy: you have a range of numbers and an array, and you have to identify which numbers are missing. Then, I was asked a question that could have been solved using k-way merge (which would have been O(n log n)), but the interviewer seemed more keen on an O(n^2) solution. I don't think they understood how k-way merge works. The second question was the edit distance for a word, which the interviewer presented dramatically (a ship captain, bad at spelling country or city names, find what he has written). I told him I had already done this question and wrote the solution in 4-5 minutes (I wrote a bottom-up procedure). He also seemed confused and didn't appear to understand exactly what I had done. The third question I struggled with a bit (it was from OS). However, time was up by then, and the interviewer said I was on the right track.

2nd Interview: (First of all: if they don't like you in the first interview, the second interview is just a formality; they'll send you back after this one.) He tried to look strict, but his questions were pretty straightforward. Some were from past projects, and one was a coding question (k linked lists merging at one point): find the first common node of all linked lists, which I solved in less than 2 minutes.

After that, they escorted me out of the building. Yeah, if interviewers don't get the solution they have in mind (even if it's inefficient), that's what happens.

Fortunately for me, I got an offer from my dream tech company (at least they understand algorithmic things and support doing things differently; I hope I could say that for Bloomberg). They should really consider increasing the level of their interviewers.

Intersection point for k linked lists Edit distance Missing numbers from 1 to n