the person

About

I design enterprise infrastructure, argue for it, and then run it.

By day I own a multi-region, multi-redundant enterprise output platform on Azure for a large global organization. It serves thousands of devices and users across four global regions, with geo-routing that sends each user to the closest region and redundancy at every layer so a regional problem doesn't take the business down. I'm the single global escalation point for the platform.

I sat in the architecture decisions for the move from on-premises into Azure. The network and firewall layers belong to other teams, and a lot of my job is holding that work to account: I found a configuration fault in one of them that was breaking DNS for the whole platform, and drove the correction with the vendor and the customer's security group. I'm also the named application owner, so production changes don't ship without my approval. I made the business case for a non-production environment that mirrors production across regions, won the funding, and defined its design, sizing, and resource requirements across three organizations. The platform teams own the provisioning.

The results I care about are the ones someone else can confirm. I brought platform-related support ticket volume down 38% year over year, measured in the customer's ServiceNow, by stabilizing the operating model and pushing documentation and self-service. I'm now redesigning the account's operational data and documentation around agentic systems and Microsoft Power Automate flows. The account's annual customer survey came back top-tier with my name on it, and I contributed to a five-year renewal.

Off the clock, I run the patterns myself. My homelab started as a way to watch movies and turned into a 9-node Kubernetes cluster on Talos Linux, managed with ArgoCD, running about fifty applications under GitOps. It all sits on a five-node Proxmox cluster with Ceph underneath. GitLab has its own VM and serves as the GitOps source, registry, CI control plane, and operating record. Runner capacity spans Linux on x86-64 and ARM64, macOS, and Windows. The macOS and Windows runners stay powered down until a job needs them. When DNS became a manual chore, I built dnsweaver as an open-source DNS project. I own its design, releases, and acceptance criteria; agents implement.

section 1

How I work

I direct AI agents to implement, and I own the engineering. Architecture, standards, integration, the decision to throw out a solution and start over, that's me. The model types faster than I do; it doesn't decide anything. I reject a lot of what gets generated, and everything ships through linting, tests, and review before it's real.

If the AI went away I'd be slower, not stuck. I've contributed features to projects like OPNsense in a language I didn't know when I started, which is the part most people skip: you still have to understand the system to direct the work well.

Each harness gets a scoped identity, so GitLab keeps the work attributable. MCP read and write access use separate tokens, and write calls require client approval. Scheduled jobs begin report-only. I decide whether verified output has earned narrowly scoped authority.

I care more about whether something works at 2am than whether it looks clever in a slide. I'd rather tell you what I've actually validated than pretend I'm certain about something I haven't.

section 2

Skills

Microsoft AzureAzure Front DoorApplication GatewayLoad BalancingCloud MigrationKubernetesDockerGitOpsGitLabArgoCDTerraform / OpenTofuAnsibleCI/CDTalos LinuxInfrastructure as CodeNetwork SecurityLinux

Recognition

  • ·Specialist of the Quarter (x2) and Specialist of the Year
  • ·Ricoh STAR Award
  • ·Microsoft Azure Fundamentals, plus multiple enterprise product certifications
section 3

Get in touch

If you're building infrastructure and you want someone who can learn a new stack fast and move quickly without shipping garbage, I'd like to hear about it.

Email → LinkedIn → Bluesky →