Awesome-Self-Evolving-RecSys

Curated list of self-evolving recommender systems: LLM-agent loops that change recommender config, code, semantic IDs or policy, with the evidence behind each gain. Companion to the survey + audit of 27 systems.

A curated list of recommender systems that improve themselves: loops in which an automated proposer, usually an LLM agent, changes a recommender's configuration, code, generative components or policy, evaluates the change, and keeps what the evaluation says is better. Each entry records what the loop changes, which signal decides what is kept, and whether code is open.