Thel Gunter
AI agents and full-stack applications in Python and TypeScript, plus computer vision on edge hardware.
- Now
- Independent software and AI engineer. Also building Inertia, sprint video analysis used at a baseball training facility, with timing gate hardware at the prototype stage.
- Recent
- July to September 2026. Built the TypeScript agent for a networking infrastructure company's device platform, now entering gradual customer rollout.
- Focus
- LLM agents, retrieval-based applications, and full-stack services, from architecture and authorization through deployment.
- Before
- Lead of software integration at Cachengo, building distributed computer vision on RK3588 edge nodes. Consulting at Capgemini.
- Base
- Myrtle Beach, SC. Remote first, open to relocation.
Selected work
- 01
Inertia
Sprint video analysis that estimates ground contact time and trunk lean from pose, used at a baseball training facility, plus a prototype of sprint timing gate hardware I designed.
Video analysis in use, hardware prototype / SAM 3D Body / RunPod serverless / ESP32-S3 / ESP-NOW
- 02
AI Agent Platform
TypeScript agent on the Anthropic API that runs device and fleet operations through an existing MCP server, with delegated per-user OAuth and confirmation gates on destructive actions.
Gradual customer rollout / TypeScript / Node / Anthropic API / MCP
- 03
Campaign Intelligence Agent
Multi-agent triage for inbound X mentions. Routine questions get a retrieval-grounded answer, judgment calls go to a human on Telegram, and posting is never a tool the model can call.
Delivered, used by the client / Multi-agent / RAG / Vector store / X API