I had one this week! Not a total "bomb", but I'm claiming recency privilege in order to talk about it anyway.
Did a full loop with a company notorious for asking Leetcode "hard" coding questions. I've been practicing "medium" questions for months, but I practiced about 10 "hards" specifically for this loop and was feeling unusually good - or at least more nervous about the behavioral interviews than the coding.
After 2 behavioral & system design interviews that went OK-ish, there was a coding interview. The first part of the question was to check if a binary tree is a valid binary search tree. I solved that part easily using the first familiar pattern that came to mind, with tests. However, the interviewer seemed confused by/skeptical of my solution, which spiked my anxiety that I had overlooked something horribly obvious or had even solved the wrong problem entirely.
For the second part, he asked me to come up with a plain text representation of a binary search tree and write code to build a tree from its text representation. This isn't even an objectively difficult problem - perhaps "Leetcode medium" at most, and it's the kind of problem I could solve alone in a quiet environment, but I have always struggled to think about recursive problems under time pressure and while "talking through my approach". I have to think hard when I tackle a recursive problem that is new to me (or even a familiar one I haven't seen in a while) - once the pattern clicks, it's dead obvious, but I definitely don't have the kind of brain that finds these problems easy. And in an interview situation, it's as if the talking part of the brain suppresses whatever other part is somewhat capable of grokking recursive problems. Particularly when there is a coding interview immediately following several very "talky" design & behavioral interviews.
Anyway, I flopped around for a while trying to get it right, but quickly ran out of time. The interviewer tried to offer guidance a few times, but this mostly just broke my train of thought, unfortunately.
The interviewer was a soft-spoken guy with an Architect job title, and my anxious, stereotyping brain, insecure about its own algorithmic ability, decided that this is the kind of engineer who can solve Leetcode hards in his sleep, so he's going to consider me extra dumb and incompetent for messing up this pretty "easy" problem. A talker who can't actually code, even. Of course, it's possible he sucks at solving algorithm problems under pressure almost as much as I do :-)
Waiting for the result.