Our EvoGit wins First Place in AgentX International Competition

We are thrilled to announce that our project EvoGit has won First Place globally in the Multi-Agent Systems track of the prestigious AgentX International Competition. Organized by the University of California, Berkeley and supported by leading tech companies such as Google, Amazon, and Hugging Face, AgentX is one of the most influential global platforms for advancing Large Language Model (LLM) agent technologies.

Overview of the AgentX - LLM Agents MOOC Competition (Spring 2025)

The competition drew nearly 1,000 teams from over 100 countries and 800 universities worldwide, including top institutions such as UC Berkeley, Stanford, and CMU. After multiple rounds of rigorous evaluation by a judging panel of 28 experts from Google DeepMind, Meta FAIR, NVIDIA, and other world-leading organizations, only 15 teams advanced to the final on-site Demo Day.

Our team from PolyU’s Department of Data Science and Artificial Intelligence (DSAI) was the only representative from China (including Hong Kong, Macau, and Taiwan) to reach the finals—and ultimately secured First Place in the most technically challenging category with their pioneering project, EvoGit.

EvoGit constructs the evolutionary process of code as a Directed Acyclic Graph (DAG), where multiple agents collaboratively evolve and continuously develop.

EvoGit is a decentralized multi-agent framework that reimagines software development as a continuously evolving process rather than a static task. It models the development lifecycle as a Directed Acyclic Graph (DAG) of Git commits, where multiple agents independently propose incremental code changes or structural crossovers and collaborate without relying on centralized schedulers or explicit communication protocols.

This architecture enables self-organizing, scalable, and resilient agent behavior. By building on standard Git infrastructure, EvoGit also facilitates seamless human-agent collaboration, allowing developers to inspect, intervene, and co-create using familiar version control tools.

EvoGit represents a significant step toward building evolvable LLM agents—intelligent systems capable of autonomously improving their own structure and behavior over time.