jobs

Research Engineer, Code RL (Reinforcement Learning)

Anthropic’s Code Reinforcement Learning team works to improve models’ ability to write, edit, test, debug, and deliver software using real tools and codebases. This research engineering position combines experimental research with hands-on

Company

Anthropic

Location

San Francisco, CA; New York City, NY

Status

Closed

Posted

2026-11-02T03:34:26.793+00:00

Anthropic is hiring a research engineer for its Code Reinforcement Learning team, which studies how models write, edit, test and debug software with real tools and codebases. The work includes designing RL environments and coding tasks, building reward signals and verifiers for code quality, running model training experiments, and improving experiment pipelines. The posting asks for strong Python, including asynchronous or concurrent programming, and end-to-end debugging. It also mentions RL, RLHF, post-training, coding agents, evaluation harnesses, program analysis, PyTorch, distributed training or GPU/TPU optimization. Hybrid; at least 25% office attendance.