# Research Engineer, Performance RL (Reinforcement Learning)  at Anthropic

> Closed role observed by RunReq as of 2026-10-15T00:28:46.484527+00:00.

## Role details

- Company: [Anthropic](http://runreq.com/company/anthropic/index.md)
- Location: San Francisco, CA
- Team: AI Research & Engineering
- Compensation: $350K–$850K /YR
- First observed: 2026-06-22T04:37:26.399850+00:00
- Days open: 94
- Employer posting: [View the authoritative posting](https://job-boards.greenhouse.io/anthropic/jobs/5160330008)

## Shortened description

### Key responsibilities

- Design and implement reinforcement learning environments and evaluations for models that write correct, fast accelerator code.
- Run experiments, guide the research plan, and deliver results into training runs.
- Work with researchers, engineers, and performance engineering specialists across and outside the company.

### Qualifications needed

- Bachelor’s degree or equivalent education, training, or experience in a relevant field.
- Expertise with accelerators such as CUDA, ROCm, Triton, or Pallas, and machine learning frameworks such as JAX or PyTorch.
- Experience across kernels, model code, and distributed systems, with a balance of research and engineering work.
- Preferred: Experience with reinforcement learning, accelerator workload porting, or large language model training methods.

## Hiring context

- Open roles on this team: 66
- Newly observed on this team in 30 days: 7
- Company median days open: 55

## Interpretation

- RunReq records observed job postings; it does not show applicant volume or completed hires.
- The linked employer posting is authoritative for the full description and application process.
- Human-readable job page: http://runreq.com/jobs/0a25f78e-5eee-8be1-a659-41c031593f0e
- Site-wide agent directory: http://runreq.com/llms.txt