I built a five-stage AI Video Workflow that handles product selection, script generation, critic review, video project planning, and operator-approved publishing. Typed schemas at each agent boundary, persisted state, and recovery-oriented tests help keep the workflow verifiable; publishing requires explicit human authorization.
In my independent AI engineering work, I also contributed deadline-time handling, focused tests, and validation documentation to DocFlow-AI. My other projects include auditable data workflows with checkpoint and resume, deterministic validation, and safeguards against unverified findings.
At Shenzhen Yuanding Intelligent Innovation Co., Ltd., I owned functional, performance, stability, and security testing across the product lifecycle, and built automation frameworks to replace repetitive manual verification. Earlier, at Jiangsu Furui Micro-Nano Sensing Technology Co., Ltd., I wrote test scripts, analyzed structured test data, and traced root causes through Linux service logs.

