Algorithm Engineer & Full-Stack Developer. Currently building SeedWorld — an emergent world simulation engine where LLM-driven characters write their own stories.
An emergent world simulation engine. Describe a world, the engine generates characters with personalities, goals, and grudges — then lets them loose. 12 mechanism systems (reputation, resources, tension, memes...) form LLM feedback loops. Betrayals, alliances, negotiations — all emergent, never scripted.
Cross-platform AI collaborative workspace with transparent execution tracing and intelligent assistant features.
Personal knowledge atomization & LLM augmented memory system designed to empower AI Agents.
The core idea: give an LLM a world description, generate characters with desires and grudges, and let them interact freely. Stories emerge from chaos — no scripts, no behavior trees. 12 interlocking mechanism systems create realistic social dynamics through LLM feedback loops.