Customer-facing execution, product strategy, AI-assisted SDLC, and hands-on prototyping — turning ambiguous ideas into working systems and helping teams adopt them.
Six prototypes built end-to-end. Each one isolates a single product, infrastructure, or workflow question — and answers it with working software.
Nine production-grade Edge/Cloud AI subapps under one hub — PPE Check, ClearRoute, Kalay CamCheck, PawNanny, and more.
Lightweight AI VMS connecting cloud, an embedded Linux gateway, cameras, mobile app, and an AI audit log.
Video intelligence sandbox — LLMs analyze 200+ video buckets, with prompt-pack tuning and judge-comparator evaluation.
Agent platform connecting persona + long-term relationship-memory modeling to LINE and Telegram channels.
Multi-day travel itinerary on a world map — climate, flights, mobile companion, all in one canvas.
Internal chatbot that turns Confluence knowledge into conversational answers with cited sources.
/ from silicon to agents
I use AI tooling to reduce the distance between product thinking and working software — not to replace engineering, but to make discovery faster, more concrete, and easier for teams to evaluate.
Clarify the user, the workflow, the constraint, and the business value. Write it down before opening any tool.
Claude Code, Cursor, agent runtimes, AI-assisted SDLC — compress idea → spec → working code into days, not quarters.
Face-to-face with sales and customers — turn live signal into the next iteration fast, and hand the artifact to the team so the loop becomes theirs.
Open to senior IC and player-coach roles where AI meets the physical world — vision AI, edge-to-cloud platforms, enterprise AI deployment · Taipei.