Anthropic Fellows Program, ML Systems & Reinforcement Learning

Anthropic Fellows Program is designed to foster AI research and engineering talent by providing funding and mentorship to promising technical talent regardless of previous experience. Fellows primarily use external infrastructure such as open-source models and public APIs to work on empirical projects aligned with Anthropic's research priorities, aiming to produce public outputs like paper submissions. The program runs multiple cohorts yearly with rolling application reviews, expecting a start in January 2027 but flexible start dates can be requested. The program includes 4 months of full-time research, direct mentorship from Anthropic researchers, access to a shared workspace, connection to the AI safety and security research community, a weekly stipend of 3,850 USD / 2,310 GBP / 4,300 CAD plus benefits varying by country, and funding for compute (~$15k/month) and other research expenses. The interview process involves an initial application and reference check, technical assessments and interviews, and a research discussion. Applicants are encouraged to apply even if they do not meet every qualification, with emphasis on diversity and inclusion. There are two main fellow tracks: ML Systems & Performance Fellows: - Mentors include Alwin Peng and Zygi Straznickas. - Projects may involve building CPU simulators for accelerator workloads, adding backends for accelerators on open source projects, building on-demand infrastructure, and creating complex synthetic data or environment pipelines. - Candidates should have strong software engineering skills, experience with complex ML systems, ability to balance research and engineering rigor, comfort with large-scale distributed systems and high-performance computing, experience with training or evaluating large language models, and skills in analyzing and debugging model training. - Fluency in Python and full-time availability are required. Reinforcement Learning Fellows: - Mentors include Ruhua Jiang, Kaidi Cao, Sunny Duan, David Brandfonbrener, Colt Steele, Dino Distefano, and Will Williams. - Projects may include building model-based tools to understand and improve AI training data, research on generalization, creating RL environments to improve Claude models and safety-related tasks, and research and implementation of RL algorithms. - Candidate criteria are similar to ML Systems Fellows. Logistics: - Participants must have work authorization and be located in the US, UK, or Canada during the program. - Shared workspaces are available in San Francisco; remote participation is allowed in the UK, US, or Canada. - Visa sponsorship is not provided; candidates must have or obtain full-time work authorization independently. - The program duration is 4 months full-time, with possible extensions and flexibility for partial commitments. Additional notes: - No guarantee of full-time offers post-program, though strong performers may receive offers or continue impactful AI safety work elsewhere. - Anthropic's full-time role policies do not apply to the Fellows Program. - The company emphasizes collaborative, large-scale AI research with a focus on safety, interpretability, and steerability. - Candidates must be fluent in Python and able to commit full-time. - The program values diversity and encourages applications from underrepresented groups. - Safety advice is provided regarding recruitment communications. Apply at the provided link.