Gittins

Gittins is an opinionated, highly optimized contextual bandit engine designed for production environments. It supports online learning with dynamic actions and context, handles non-stationarity, and enables safe reward handling with efficient algorithms. Implemented in Rust with Python and WebAssembly bindings, it provides fast, deterministic decisions and supports offline policy evaluation via an appended experience log. It targets developers and data scientists implementing contextual bandit algorithms for adaptive decision-making in varied applications, such as personalized recommendations and dynamic content selection.