ARENA is a 4-5 week intensive ML engineering bootcamp in London that trains technically skilled individuals to contribute to AI safety research. It covers deep learning fundamentals, mechanistic interpretability, reinforcement learning, and model evaluations.
ARENA is a 4-5 week intensive ML engineering bootcamp in London that trains technically skilled individuals to contribute to AI safety research. It covers deep learning fundamentals, mechanistic interpretability, reinforcement learning, and model evaluations.
People– no linked people
Updated 04/02/26Funding Details
Updated 04/02/26- Annual Budget
- -
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- -
Org Details
Updated 04/02/26ARENA (Alignment Research Engineer Accelerator) is a technical upskilling programme founded in 2022 by Callum McDougall with the aim of growing the pipeline of ML engineers working on AI safety. The programme is delivered as a 4-5 week intensive in-person bootcamp, currently hosted at the London Initiative for Safe AI (LISA) in Shoreditch, London. ARENA is a programme of LISA (registered UK charity Safe AI London Ltd, number 1211693). The curriculum is structured around four core technical areas: deep learning fundamentals, transformers and mechanistic interpretability, reinforcement learning, and LLM evaluations. Participants work through structured pair-programming exercises and then transition to open-ended capstone projects with teaching assistant supervision. The programme covers accommodation, meals, and travel expenses for all participants to remove financial barriers. Callum McDougall ran the first three iterations of ARENA and has since taken an advisory role, now working as a research scientist on DeepMind's interpretability team. The programme is led by James Hindmarch (Programme Lead) and Joly Scriven (Operations Lead), with James Fox as Director, Chloe Li as Strategy and Curriculum Developer, David Quarel as Head TA, and a rotating team of specialist teaching assistants. As of early 2026, ARENA has completed seven cohorts and is preparing its eighth (ARENA 8.0, May-June 2026). Cohort sizes have grown from approximately 16 participants in ARENA 2.0 (2023) to around 30 in recent iterations. Alumni have gone on to roles at Apollo Research, Anthropic, METR, OpenAI, and the UK AI Safety Institute, become MATS scholars and LASR participants, and started their own AI safety organisations. ARENA also publishes its curriculum openly at learn.arena.education, which is used by independent learners and other programmes globally including the Finnish Alignment Engineering Bootcamp and the Technical Alignment Research Accelerator (TARA) in APAC. The programme is funded by Coefficient Giving and has received grants through Open Philanthropy's AI safety capacity-building programme.
Theory of Change
Updated 04/02/26ARENA believes that a primary bottleneck to AI safety progress is the shortage of ML engineers with the technical depth and safety orientation needed to do high-quality alignment research. By running intensive bootcamps that combine structured technical training with hands-on project work and immersion in the AI safety research community, ARENA aims to rapidly move talented engineers into productive safety roles at leading organisations. Each cohort graduate who joins a safety-focused team or starts independent research represents a direct addition to the field's capacity. The free online curriculum multiplies this effect by enabling self-study and spin-off programmes worldwide.
Grants Received
Updated 04/02/26Projects– no linked projects
Updated 04/02/26Discussion
No comments yet. Be the first to share your thoughts.