From cables to agents, I make the whole system hold up.
The work now is AI delivery in classified and constrained environments.
// how the work holds
SLOP → TRUSTRaw model output is cheap and wrong somewhere. The Construct is the gates that make a subset safe to ship.
THE WORK
Observe every step · validate with a model from a different vendor family · gate before execution · design the recovery path. The agent that wrote the code never grades its own work.
// what I operate across
SUBSTRATE → PLATFORM → FRONTIEROne through-line across three layers: you can see the system, reverse a bad change, and promote without a hero.
- 01BUILD THE SUBSTRATE
Infrastructure Engineering
Bare metal, networks, storage, and a 100+ GPU fleet at the tactical edge. The substrate the rest of the stack sits on.
- 02PAVE THE PLATFORM
Platform Engineering & Distributed Systems
Kubernetes and GitOps across 100+ clusters: promotion, rollback, and artifact traceability other engineers run without calling the person who wrote it.
- 03OPERATE THE FRONTIER
AI Infrastructure & Multi-Agent Systems
Air-gapped LLM inference, AgentOps validation gates, and a Kubernetes control plane for multi-agent work. Observe, validate with a different model family, gate, recover.
// explore
PICK A ROOM →If the problem is infrastructure, platforms, or AI that has to ship under hard constraints, write me.