Multi-Agent Deep RL meets Darwinian evolution: predator/prey agents learn under sparse reproduction-only rewards (no shaping, no cooperation signal) while heritable traits mutate and are selected across generations — a live testbed for the Baldwin effect, and for whether cooperation, defection, and free-riding emerge unassisted.

[ACL 2025] Official respository for EvoPatient: LLMs Can Simulate Standardized Patients via Agent Coevolution

MatrixWorld: A pursuit-evasion platform for safe multi-agent coordination and autocurricula

Multilingual Code Co-Evolution Using Large Language Models

Adversarial co-evolution orchestrator: an executor LLM improves an artifact, a deterministic scorer judges it (keep-if-better via git), a validator LLM advises — until quality peaks. Off-the-shelf agent CLIs, walk-forward scoring, live web dashboard. General-purpose, not just trading.

Multi-provider AI routing engine with decision quality scoring. Routes queries to Claude/GPT-4/Gemini based on task type, cost, and performance. Self-improving via pattern analysis.

University Evolutionary Computing project exploring how DE, CMA-ES, genetic algorithms, and co-evolution can optimize simulated robot bodies and neural controllers in EvoGym to maximize performance across diverse environments.