Anthropic Fellows Program, ML Systems & Reinforcement Learning

Anthropic's mission is to create reliable, interpretable, and steerable AI systems. The Anthropic Fellows Program is designed to foster AI research and engineering talent, providing funding and mentorship to promising technical talent regardless of previous experience. Fellows work on an empirical project aligned with Anthropic's research priorities, primarily using external infrastructure (e.g., open-source models, public APIs), with the goal of producing a public output such as a paper submission. The program runs for 4 months full-time, with multiple cohorts per year and rolling applications. The next cohort is expected to start in January 2027.

• 4 months of full-time research
• Direct mentorship from Anthropic researchers
• Access to a shared workspace in Berkeley, California or London, UK
• Connection to the broader AI safety and security research community
• Weekly stipend of 3,850 USD / 2,310 GBP / 4,300 CAD plus benefits (vary by country)
• Funding for compute (~$15k/month) and other research expenses

This posting covers two workstreams: ML Systems & Performance and Reinforcement Learning.

Projects may include building a CPU simulator for accelerator workloads, adding backends for different accelerators on an open source project, building on-demand infrastructure for other infrastructure-heavy fellows projects, and building complex synthetic data or environment pipelines.

Projects may include building model-based tools to better understand AI training data and improve training data quality, researching generalization, creating RL environments to improve Claude models at capabilities within your domain of expertise, building RL environments for safety-related tasks, and conducting research and implementing solutions in areas such as RL algorithms.

Ideal candidates have strong software engineering skills with experience building complex ML systems, can balance research exploration with engineering rigor and operational reliability, enjoy collaborating across research and engineering disciplines, are comfortable with large-scale distributed systems and high-performance computing, have experience with training, fine-tuning, or evaluating large language models, and are adept at analyzing and debugging model training processes. Candidates must be fluent in Python and available to work full-time on the program.

To participate, you must have work authorization in the US, UK, or Canada and be located in that country during the program. Workspaces are available in London and Berkeley, and remote fellows in the UK, US, or Canada are also considered. Visa sponsorship is not available for fellows. The program runs for 4 months, full-time.

Apply at the bottom of this page. Applications are reviewed on a rolling basis. The interview process includes an initial application & reference check, technical assessments & interviews, and a research discussion.