I build production AI agent systems
and the infrastructure they run on.
Consulting and engineering for teams shipping LLM-powered products. From tool calling and orchestration to the Kubernetes stack underneath — I work end-to-end on systems that have to actually run, not just demo.
Featured work
ORION
Multi-agent orchestration platform
Web app + worker that routes messages between Claude, Ollama, and OpenAI agents, with native MCP tool calling, GitOps integration, and a Talos K8s control plane.
Orion's Belt
Local LLM agent — desktop distributable
Flask app that runs locally with PII guard, LanceDB vector memory, and MCP tools — the offline counterpart to ORION. PyInstaller-packaged for offline distribution.
Homelab
The platform that runs the rest
Talos Kubernetes cluster with Gitea, Authentik SSO, CrowdSec, ArgoCD-style GitOps, and a monitoring stack. The site you're reading is served from it.
Have an LLM project that's stuck at the "demo works, prod doesn't" stage?
That's exactly the gap I help close — turning prototypes into systems with proper tool calling, evaluation, observability, and cost discipline.
→ start a conversation