The interview process involved a recruiter call, a 60-minute technical screen, a hiring manager call, and finally an on-site loop with more coding and some behavioral questions. I did not make it past the technical screen.

The technical screen was a pre-made coding problem in Google Colab. It differed from what the recruiter had prepared me for. I was informed it would be a practical, non-LeetCode style coding exercise in my chosen environment and language, involving concurrency/multithreading, and that communication with the interviewer would be most important.

The interviewer sent me a pre-made Colab notebook with a problem to convert nested stack traces into discrete start and end events. They provided a small amount of explanation at the beginning. There was no concurrency involved, and it seemed they perhaps expected a solution using recursion.

Very little feedback or guidance was given throughout the interview. The interviewer appeared either inexperienced/unprepared in interviewing or very disengaged and uninterested in conducting a technical screen. The interviewer asked no questions of me at any point, beyond adding a specific edge case. At the end, I had to ask about their role at Anthropic, their team, and how long they had been there, as no time was allocated for introductions. They were not from the team I was interviewing for and did not seem familiar with that team.

While not the worst technical screen I've experienced, it was overall sub-par and disappointing for a company seeking to recruit experienced talent. There was clearly a disconnect between recruiting and engineering regarding the technical screen content, as it appears they have multiple versions. Additionally, the interviewers need training on conducting effective interviews.

It is also worth noting that Anthropic's policy is not to provide feedback on interviews.

Given a list of samples, where each sample contains a list of function names, generate start and end events for each function across the list of samples.