Recipe-driven recommender systems on irspack — one YAML = one model = one recommendation API. Train writes a signed artifact; serve mounts it with hot-swap. Single binary, no database.
python recommendations machine-learning personalization collaborative-filtering matrix-factorization bpr recommender recommendation-system recsys recommendation-engine recommender-system implicit-feedback serving mlops fastapi optuna ials recommendation-api irspack
-
Updated
Jul 11, 2026 - Python