Iām Elliot. I enjoy creating tools for machine learning research, particularly Reinforcement Learning.
My personal interests include photography (see my portfolio site), travel, skiing, and music (last.fm, spotify). I previously competed professionally in eSports, and now volunteer as a photographer for in-person events (see my blog).
š¤ Recent project: implementing two of my favorite board games and training RL agents: gobblet-rl and cathedral-rl.
š„ A live version of Gobblet running in the browser can be viewed here (using WebAssembly, see my tutorial for details).
š I am the Project manager for the PettingZoo library, the standard API for Multi-Agent Reinforcement Learning (MARL).
To learn more, see Announcing the Farama Foundation and the Farama Team.
š PettingZoo 1.23.0 is now live! (release notes) This release standardizes the API to fully match Gymnasium, and includes many bugfixes, pickling support, and a documentation overhaul. Released alongside is SuperSuit 3.8.0 (release notes)
š I am excited to announce a mature release of Shimmy, an API conversion tool for using popular external RL environments with Gymnasium and PettingZoo (blog post) (release notes)
ā
This website was created using GitHub Pages, with workflow actions to fetch WakaTime stats, recent activity, and render a preview image of my resume PDF. The source code can be found here: https://github.com/elliottower/elliottower.github.io