The OA is kinda funny for a Java company. They asked two hardish HackerRank questions in a two-hour interview format. The first was simply implementing an in-memory cache (why would you do this in Java?). However, the question was so vague and did not include a single example other than a data contract. I think I spent 30 minutes understanding the problem and less than 15 minutes writing code for it. The question goes entirely against Java architecture anyway, and they provided feedback that Redis cannot be used as an LRU cache when I commented that would be my production approach.

They provided broken boilerplate code that seems like it was plugged into ChatGPT and prompted, "Convert this interview question from .ts to Java" for this first interview question. Maybe I'm just dumb, but I think anyone who can code a modified Bellman-Ford with rounding edges in time without cheating, on top of figuring out what the easy OOD question is even asking, probably can work wherever they choose to.

Modified Bellman-Ford dynamic programming.

This is a grad-level algo.