The recruiter reached me on LinkedIn. We set up a phone screen, which was more about the position.

Then I had a technical coding interview on CoderPad with another data engineer. It was not hard at all, with all basic questions around strings, lists, and dictionaries. I answered five SQL and five Python questions, totaling one hour.

Finally, I had my on-site interview. For some reason, the recruiter called them "full-stack questions," but there was no CSS or JavaScript coding involved. All on-site questions were regarding DW design, table design, schema design, and so on. The interviewers asked SQL questions based on the DW that I designed for them.

The on-site questions were no match for the CoderPad interview. You need to have a good understanding of system architecture and system design.

I personally could not finish one Python question. The question was about updating metrics (I had defined the metrics to answer an earlier question) for a Facebook module as a stream of new data was coming in. As the interviewer pointed out, I was on the right track by having a session ID and user ID to de-duplicate the same user activity, but it was not enough, I guess.

My impression was that they want someone who answers quickly and smoothly in all questions. I literally just missed 25% of a Python question, and I got rejected. I also received feedback that I have good product sense and coding experience, which made me feel like they expected me to answer all of their questions.

One last note: They ask questions to ensure that you are not using the data engineer position to switch to something else, like data scientist. I heard a friend aced the technical but was rejected because of that in the last round.

Flatten a nested dictionary?

(This was the hardest question in CoderPad.)