I build practical AI-native products: desktop agents, voice interfaces, enterprise knowledge systems, and full-stack agent applications.
I work at the layer where AI becomes product:
- voice-to-action tools that fit into real desktop workflows
- RAG and enterprise knowledge systems for teams
- agent-powered applications with identity, matching, chat, and user models
- rapid MVPs that can become release-ready software
My standard is simple: a product should feel useful in the first session, not only impressive in a demo.
| Product | Role | What it proves |
|---|---|---|
| TypeUp | Founder / builder | Windows desktop AI voice transcription and editing client with local engine orchestration, Electron UI, Python agent, and release packaging. |
| Enterprise KB | Full-stack AI product builder | Enterprise RAG shell around RAGFlow: users, permissions, document workflows, streaming AI chat, training modules, and admin operations. |
| Second-Me Skills / Qiubi | Agent app builder | Full-stack SecondMe / agent application work, including social product flows, OAuth, daily matching, chat, community, and reusable build skills. |
| VoxType | AI interaction prototyper | Voice-input and AI typing experiments for practical productivity workflows. |
I am building in the AI product space with a bias toward shipping.
Product judgment
I care about real user workflows, not model novelty alone. I think through the full loop: problem, interface, data, model behavior, feedback, packaging, and iteration.
Engineering depth
I can move from frontend to backend to desktop runtime. I am comfortable building the product surface, integrating APIs, managing local engines, and preparing releases.
AI systems focus
My work spans RAG, voice, agents, desktop automation, streaming chat, OAuth, local-first tools, and knowledge workflows.
I am especially interested in:
- AI productivity tools that reduce friction in everyday work
- local desktop agents that make model output immediately actionable
- enterprise RAG systems with real permissions, operations, and workflow design
- agent-native applications that combine identity, context, chat, and decision support