The same job, one layer up each time.
// the short version
I understand the stack because I came up through it.
A decade of it: learn what the software depends on, make the whole system hold, and leave the people standing there able to run it without you. From cables and hospital networks to classified infrastructure, distributed platforms, air-gapped AI, and multi-agent systems.
// the arc
Started at the help desk, turning up cables and troubleshooting the systems behind them. Self-taught Linux, networks, and hardware. The credentials came after the practice, earned while working. That order never reversed.
Built and managed 40+ healthcare networks across Virginia under full HIPAA compliance. The first arena where being wrong had consequences.
The actual job title was Forward-Deployed Engineer, at GDIT. Stood up classified networks and led a 0-to-1 datacenter build. Wrote my first production Python automation. The teaching thread starts here too: daily technical briefings for senior non-specialist leaders. Complex systems, plain words, every day for three years.
At GDIT/NGA, built air-gapped LLM infrastructure on a 100+ GPU fleet. Release automation in Python cut deployment rework ~80% across 100+ Kubernetes clusters. Declarative GitOps for 58 applications added promotion, automated rollback, and artifact traceability. Provisioning cycles went from weeks to hours. The teaching thread continued through mentoring and onboarding that cut ramp-up from ~3 months to ~1 week.
AgentOps, an open-source operating loop for coding agents in Go: one intent, one bounded build, one fresh verdict, and portable skills, 400+ GitHub stars. Gastown Operator: controller-runtime and CRDs for multi-agent workspace orchestration. An air-gapped AI platform reference, 139 Helm templates deep, the full inference and agent stack. Ecosystem contributions: Gastown, Beads, KAgent. 20+ articles at /writing.
Staff Platform Engineer in defense autonomy. Bare-metal Kubernetes across air-gapped and restricted environments; GitOps promotion pipelines. Make it boring to promote a tested change across those boundaries, with reliability designed into the platform and the handoff.
// the loop
THE LOOP — Cables → systems → networks → infrastructure → distributed platforms → AI → agents. Every chapter, the same move one layer up: understand the dependencies, take the frontier where the mission is, make it hold, and teach the people standing there.
The case studies are at /work, each one stated as problem, constraint, what I built, and the number. The beliefs behind them are at /method.
// education
B.S. Information Technology. M.S. Cybersecurity (WGU). Practice first, paper confirming.
// how to reach me
Got a system that has to hold under hard constraints? Write me about the system and the constraints. GitHub has the public work; LinkedIn has the longer trail.