Research Engineer, Cybersecurity RL (Reinforcement Learning)

Research Engineer on the Cybersecurity RL team within Horizons, the group that leads reinforcement learning research for Anthropic's Claude models. The team works on safely improving model ability in secure coding, vulnerability remediation and other defensive security tasks. The engineer designs and builds RL environments, runs experiments and evaluations, puts the results into production training runs, and collaborates with researchers, engineers and security specialists inside and outside the company.

Anthropic wants cybersecurity research experience combined with machine learning and strong software engineering, along with the ability to move between exploration and implementation. Security engineering, fuzzing or detection and response experience, CTF or cyber range work, and familiarity with RL and LLM training help. Based in San Francisco or New York.