Anthropic Interview Questions

Recruiter reached out and they contacted my references right away. Only after my references cleared, they sent me an OA, which I thought was pretty difficult. I solved 3 out of 4 questions and then they rejected me.

The process is very technical and centered on “First Principles.” After the initial screen, there’s a strong focus on a “Pair Programming” session and a deep “System Architecture” dive. Different from other Big Tech loops, Anthropic expects you to reason about high-scale distributed systems with a big focus on safety and reliability. They’re not just looking for someone who can code; they want someone who understands the “Why” behind every architectural decision.

1. Recruiter Screen 2. Hiring Manager Call about motivations and experience 3. Agents Coding Screen - modify an existing agent loop to add more tool calls, chat about optimisation 4. Roleplay/ML System Design with a pretend client 5. Bug Fixing - Python, notebook, open book, fix bugs 6. Culture Fit / Behavioural The technical bar is quite low/basic. The interviewer quality feels inconsistent. The bug fixing seems to probe random, obscure concepts. Meanwhile the agents coding is extremely simple and the system design is almost non-technical. Strange mixed bag. The interviewers the whole way seem to be scoring and hunting for some hidden rubric/agenda.