Anthropic Software Engineer interview experience

In one phone-style coding interview, I worked on a “find duplicates” file problem where I had to implement the file-system logic myself—no helper scaffolding. I was allowed to use Google, but even with that, the time went quickly because the bulk of the work was figuring out a reliable way to hash or compare content efficiently. After the main coding work, the interviewer pushed on potential failure points around hash behavior and how I’d improve efficiency, including thinking about distributed approaches.

Across the process, I also hit a separate LRU cache challenge where I was given a