Zimu Wang

Zimu Wang

UC Berkeley CS undergraduate building self-improving and reliable AI agents.

Projects

These projects are part of Simple Agent Lab, an open-source research collective exploring simple yet effective methods for AI4AI and self-improving systems.

RSIHub Permalink

GitHub stars Lead developer; drove the project from 0→1. An open-source recursive self-improvement framework that evolves agent prompts, skills, harnesses, and code while preserving frozen evaluators and auditable experiment lineage.

Simple Long Horizon Agent Permalink

Built an on-demand skill runtime and unified four agent presets behind a composable AgentSession/toolset API with config-driven MCP support.