I keep returning to one question: how can technology help ordinary people understand complex knowledge, organize work, and reach more opportunities? That concern has drawn my work toward AI for Science, medical AI, and agent application development: systems that make knowledge work more legible, collaborative, and deliverable.

Much of my work sits between field reality and working AI products: user interviews, workflow abstraction, front-end prototypes, agent tool orchestration, RAG evidence chains, and structured reports. Recent projects include terminal medical record quality control, rare disease analysis agents, AI Scientist research workspace, and tabular prediction.

Beyond product work, I write on Douban for ordinary women navigating career transitions and technical learning. 2k+ followers, 400k+ reads; mostly, a small attempt to make useful information less exclusive.