Research Engineer, Finetuning
- Location
- San Francisco, CA, United States
- Track
- AI Researcher
- Posted
- November 21, 2026
- Source
- Greenhouse
Job description
About the role
We're looking for a Research Engineer to help make Claude more capable, reliable and useful on complex, real-world work. You'll work across research and engineering, from early ideas and experiments through to improvements in our models.
This is an early-stage team working on open-ended problems. Priorities will change as we learn, and you'll have a lot of ownership over what you work on. We're looking for someone who's excited about that, and who enjoys working closely with both research and product.
Key responsibilities
- Design and run experiments to improve model capabilities
- Build reinforcement learning environments and evaluations
- Develop the tools and infrastructure needed to support this research
- Collaborate with teams across research, engineering and product
- Help shape the direction of a new area of research
Minimum qualifications
- Strong software engineering skills, including proficiency in Python
- Experience training, fine-tuning or evaluating machine learning models
- A track record of owning ambiguous technical projects end to end
- Strong communication skills and a collaborative working style
- Care about the societal impacts and safety of your work
Preferred qualifications
- Experience with reinforcement learning, particularly in long-horizon or sparse-reward settings
- Experience building reinforcement learning environments, simulation systems or large-scale ML infrastructure
- Experience building evaluations or benchmarks for agentic systems
- Experience working closely with product teams to drive model improvements
Representative projects
- Build a simulation where decisions made early on have consequences much later, and use it to train a model
- Design an evaluation for work that takes a long time and has no single right answer
- Investigate where a model's independence breaks down on long projects, and run experiments to improve it
- Develop a way for people to oversee a large amount of work without reviewing every step
Annual Salary
$500,000 — $850,000 USD