Experience

Main work

• Led 0→1 medical AI knowledge apps: designed and built a pluggable RAG / KAG runtime and medical ontology knowledge base for the IHS clinical line.
• Contributed to AgentCC / AGT, wiring knowledge retrieval, task orchestration, and business actions into a usable loop.
• Established automated gates for style, architecture, file size, and color—tightening exceptions over time. 479 commits in ~6 weeks (28.7% of team commits).

Main work

• Owned SQL pre-risk prediction and high-load incident-response agent scenarios end to end: requirements, PRDs, prototypes, interaction design, QA, integration, and launch.
• Designed staged execution and reply control for diagnosis failures (context loss, unstable steps, off-topic answers) via roles, templates, upfront rules, output constraints, and fallbacks.
• Drove cross-functional delivery (reviews, UI iteration, FE/BE integration); flagged P0 blockers; built a mock frontend when UI lagged; structured Alibaba / Microsoft competitor analysis.

Main work

• In a 3-person team covered product, engineering, content, production, and fulfillment; 20+ orders on a peak day without paid ads; RMB 20,000+ monthly sales at steady state.
• Built Xiaohongshu growth workflows: ranking collection, trend breakdown, topics, text/image generation, batch editing, and review loops.
• Drove cold-chain → ambient SKU shift after feedback, competitor, and cost analysis; ~60% per-order margin.

Main work

• Built Year-2 event systems (admin, applications, submissions, judging, hotels) for 800+ participants and 200+ teams.
• Paired core engineering with PM work; refactored application tables toward search, preview, multi-select, and fuzzy-match flows.
• Helped resolve on-site DB mistakes, traffic spikes, DDoS, and login failures via rollback, Cloudflare hardening, degradation, and rapid fixes.