Retriever Controlling Systems
OngoingA React PWA that controls the field retrievers — fast and responsive, communicating over WebSocket and MQTT for low-latency feedback.
- React
- PWA
- WebSocket
- MQTT
whoami
Fullstack engineer — web, embedded, and AI
I design, build, and run production systems end to end — web frontends, Node.js and Python backends, data architecture, and the embedded firmware that talks to hardware in the field. Most of what I ship, I own alone: from first commit to the thing running unattended.
01 · range
Nine production systems, designed and maintained alone — architecture, implementation, and the on-call pager when something breaks.
The same week can mean debugging an ESP32 interrupt handler in C++ and shipping a React feature — with the Node.js and Python services in between built by the same hands.
Game dev and mobile app roots before this — Android, iOS, Windows, Linux. AI work follows the same pattern: self-hosted models, not API wrappers.
02 · work
A React PWA that controls the field retrievers — fast and responsive, communicating over WebSocket and MQTT for low-latency feedback.
A React PWA that controls the turning targets in the field.
A master controller for coordinating multiple retriever units and building scenarios through a scenario editor.
A fleet-management dashboard: live status for every device in the field, remote debugging and log access, and historical status over time.
End-to-end chatbot built on a self-hosted Ollama deployment — model serving, a thin API layer, and a chat UI, with no third-party inference API in the loop. Feedback from conversations feeds back into the index to improve retrieval over time.
03 · skills