Developer Relations · Open-source Ecosystems · AI Agent Evaluation
I have led open-source developer programs, worked as a FastDeploy PM, contributed engineering fixes to PaddlePaddle projects, and built evaluation tasks and infrastructure for coding and web Agents.
- PaddlePaddle developer ecosystem — Led two open-source hackathons and worked on contributor programs, university activities, documentation, GitHub workflows, and community metrics. Contributed merged fixes and documentation to PaddleOCR and PaddleSpeech.
- FastDeploy PM — Worked on developer adoption, release support, documentation, and user feedback. Identified V100/SM70 compatibility as an adoption blocker, then implemented and validated the engineering adaptation and contributed related CI fixes.
- Agent and model evaluation — Led Code Editing/IRE and core RiskChainBench work; collaborated on MAB-Bench and DF-Bench. Designed tasks and data, built containerized evaluation environments, ran multi-model experiments, audited results, and analyzed failure trajectories.
- Open-source Agent tooling — Built Model PR Gate and contributed to DSH plugin migration and accessibility tooling.




