I've spent 25+ years in IT infrastructure — networks, security, servers, and everything it takes to keep business systems running. Somewhere along the way, containers turned into a genuine passion, and lately I've focused on what I think is the next big infrastructure challenge: running AI agents as secure, well-managed workloads.
Most teams adopting AI agents today hand them far too much access. My approach is different — agents work through containerized API gateways that hold the credentials, so the agent only ever sees narrow, scoped interfaces. Everything is versioned in Git, documented for humans and agents alike, and gated by human review where it matters. It's the same isolation and least-privilege discipline we've always applied to production systems, extended to a new workload class.
What I Do
Containers & Orchestration Docker, Docker Compose, Docker Swarm, and Kubernetes — with a particular interest in secrets management, network policy, and running AI agent runtimes safely. CKA candidate.
Security Risk assessments, system hardening, and two-factor authentication at scale (YubiKeys, WatchGuard AuthPoint, Duo). AI agent runtime security: credential isolation, deny-by-default egress, prompt-injection mitigation through architecture.
Networks & Firewalls UniFi networks and wireless, MikroTik, HP, and Ubiquiti switching; pfSense, UniFi, and WatchGuard firewalls.
Platforms Windows and Windows Server administration, Linux server administration and CLI, Microsoft 365 / Office 365 (migrations through troubleshooting), macOS/iOS/iPadOS integration into corporate environments.
Automation & Scripting PowerShell automation, Python for data manipulation and tooling, JavaScript for web interfaces, and Java contribution to enterprise backend systems.
Selected Projects
Containerized API Gateway Architecture — Docker-based gateways that isolate credentials from AI agents, letting agents work with tools like UniFi and ConnectWise Manage through scoped interfaces. Fully documented, including an AGENTS.md file so agents can onboard to the architecture. Maintained as code in Git with complete change history.
AI Agent Control Plane Rollout — Standing up an open-source, self-hosted agent management platform with deny-by-default network egress policy, RBAC, budget caps, and mandatory human review stages.
Let's Talk
If you want to discuss infrastructure, containers, network engineering, or what it takes to run AI agents without handing them the keys — reach out. I'm always up for a good technical conversation.