The interviews in theory are a good structure, but as others have said, the interviewers suck. One was a basic PR review. I finished it early, and they had no notes because I had touched upon everything plus bugs that they didn't even realise existed.

Second interview was a small LeetCode style problem involving encryption. When I started the interview, I had my shell open and was previously using Poetry to run my scripts and tried to do this from muscle memory. Then literally straight after I ran the correct Python command to run a script, the feedback was: "You did not know that you could run a Python file using the command: 'python'." I was literally marked down because of muscle memory and typing a standard command to run Python.

They also turned it into a math test when I couldn't figure out the remainder of 13/90 (not the real numbers, but a valid representation) without a calculator. I asked if I could just run my function to get the answer, and he said it would be cheating. Since when is running your own code to see output cheating? Their feedback was: "There were some issues with basic math." I asked a friend who solely studied math and works as a data scientist this question, and he got it wrong. These interviewers clearly have just looked at this problem before and know the answers.

My code worked in the allotted time, but apparently this was a math test. The interviewers from the PR review were nice though, and the structure in theory is good, just engineers who expect perfection, no typing errors, no math mistakes, and it's not important if the code works.

Review a PR and write a function. There was nothing about me.