Code RL Research Engineer — End-to-End Coding Pioneer

Neura Market; San Francisco (CA); Hybrid; USD 500,000 - 850,000; Full time

Benefits offered by this job

Equity donation matching

Generous vacation

Parental leave

Flexible hours

Office space

Job summary

Anthropic in San Francisco is hiring a Research Engineer for the Code RL team. You will advance our models' ability to write, edit, test, debug, and ship real software end-to-end on real codebases and tools.

This role blends research and engineering: you will design RL environments, create reward signals and verifiers, run training experiments on frontier models, and improve the speed and reliability of the pipelines that enable rapid iteration.

Qualifications

• Strong software engineering and Python expertise with asynchronous programming.
• Experience owning end-to-end systems and debugging across the stack.
• Ability to balance research with engineering and to design experiments.
• Commitment to code quality, testing, and performance.
• Passion for safe and beneficial AI.

Responsibilities

• Design RL environments and coding tasks for real-world codebases.
• Develop reward signals and verifiers that define what good code means.
• Run training experiments on frontier models and diagnose why they improve.
• Improve the speed and reliability of data pipelines supporting experiments.
• Align your focus with the area where you can have the greatest impact.

Skills

Python expertise

End-to-end debugging

Experiment design

Code quality

AI safety

Education

Bachelor's degree in CS or related

Tools

PyTorch

CUDA

Job description

Anthropic in San Francisco is hiring a Research Engineer for the Code RL team. You will advance our models' ability to write, edit, test, debug, and ship real software end-to-end on real codebases and tools.

This role blends research and engineering: you will design RL environments, create reward signals and verifiers, run training experiments on frontier models, and improve the speed and reliability of the pipelines that enable rapid iteration.