Anthropic
Research Engineer, Code RL (Reinforcement Learning)
About this role
Anthropic is seeking a Research Engineer to advance Claude's code generation and software engineering capabilities through reinforcement learning. You'll design RL environments, build reward systems, run training experiments, and optimize pipelines to help AI models write, test, debug, and deploy real software end-to-end.
What you'll do
- Design RL environments and coding tasks that teach models real software engineering workflows
- Build reward signals and verifiers to define what constitutes correct, high-quality code
- Run and analyze training experiments on frontier models to diagnose improvements and failures
- Debug system performance across the full stack and optimize training pipelines
- Collaborate with alignment and production teams to deploy research innovations at scale
- Focus on areas like agentic coding behaviors, code correctness, autonomous engineering, or accelerator performance
What they're looking for
- Python expertise including async/concurrent programming
- Full-stack system ownership and debugging
- Research design and rigorous experimental analysis
- Reinforcement learning, RLHF, or LLM post-training
- PyTorch and distributed training systems
- Program analysis, testing, or formal verification methods
- GPU/accelerator performance optimization
- Coding agent or sandbox development experience
Opens the application — the Jobs AI extension fills it for you. Set up autofill
Opens the official application on the employer’s site. No login required.
Anthropic
Anthropic builds Claude, an AI assistant, and is hiring for engineering roles across infrastructure, data systems, and security that support both AI research operations and the company's internal technology needs. The company seeks infrastructure engineers, systems integrators, data scientists, and security specialists to build production-scale systems for training data pipelines, financial operations, developer productivity measurement, research infrastructure, and server firmware security.
- Website
- anthropic.com
Likely interview questions
- Walk us through your experience designing reward signals or verifiers. How have you determined what 'good' looks like in a complex task?
- Tell us about a time you debugged a system across multiple layers of abstraction. How did you approach it and what tools did you use?