Product-minded engineer building AI systems, computer vision, IoT, and agentic developer tools.
I turn rough product ideas into shippable systems, from browser extensions and mobile apps to industrial edge AI platforms.
|
Real-time production tracking platforms with IoT sensors, MQTT event pipelines, operational dashboards, APIs, and databases designed for factory workflows. |
On-device detection, tracking, model conversion, embedded inference, and camera workflows using K230/RISC-V boards, YOLO-style models, VLMs, and custom datasets. |
|
Developer-facing AI tools, local translation, local LLM gateways, embeddings, RAG, speech processing, and workflow automation that stay cheap, private, and useful. |
Product-grade apps, browser extensions, dashboards, reservation flows, social layers, and mobile experiences built with clean UX and pragmatic infrastructure. |
|
Public open-source · npm A real-time multi-agent coding CLI where agents share live context, co-edit without locks, run in dedicated terminals, and support headless CI runs. |
Public open-source A local-first browser extension for translating Chinese websites without API keys, token budgets, or paid translation services. |
Private/local work Production dashboards, ESP32/RFID sensor flows, MQTT ingestion, TimescaleDB analytics, FastAPI backends, and Vue/Next.js operational UIs. |
|
Private/local work K230 edge vision, YOLO INT8 deployment, multi-object tracking, textile-state VLM experiments, auto-annotation, and MQTT event publishing. |
Private product A browser extension and web app for pixel-anchored discussions on any webpage, with realtime sync, moderation, profiles, credits, and a Pro plan. |
Private product A mobile AI companion product built around long-term personality, memory, retention, Supabase, embeddings, and async AI jobs. |
- Industrial AI for production monitoring, traceability, and operational decision-making.
- Edge computer vision that can run close to the machine, not just in the cloud.
- IoT systems that connect sensors, devices, brokers, APIs, and dashboards cleanly.
- Agentic developer tools that make multiple AI coding agents observable, steerable, and practical.
- AI products with a sane cost model: local-first when possible, cloud APIs only where they matter.
- Open-source tools that are easy to install, honest about tradeoffs, and useful on day one.
I keep this profile focused on original work and active directions. Forks, old experiments, and private repos are not featured here unless they explain what I am currently building.
Building practical AI and product systems for the physical world and the web.