Kubernetes Engineer · Senior / Staff
Teams on my clusters ship more and think about Kubernetes less.
I stood up my first production clusters in 2015 with home-brewed Ansible. The tooling has changed every year since and the job hasn’t. Build the cluster, pave the road onto it, and make deploys boring enough that the team forgets they used to be scary.
Eleven years of production clusters
Four generations of Kubernetes, from fleet in 2015 to the cluster serving my house today, each with a measurable result.
At Food Service Warehouse I ran Kubernetes on vSphere with CoreOS, fleet, etcd, and flannel. The Node.js teams on those clusters had the fastest release cycle in the company.
At CyberGRX deploys were a brittle quarterly event. I wrote a Go operator that runs blue/green rollouts as a custom resource, rebuilt the deploy process around it, and had every engineer in the org shipping with it inside 3 days. From then on they deployed whenever they wanted.
At Cloaked I led the migration from a legacy PaaS to AWS EKS, with a multi-account model that met SOC 2, ISO 27001, and ISO 27701 requirements, and a GitOps CI/CD pipeline that accelerated deployments 30x.
My homelab is a three-node k3s cluster on M4 Mac minis. Flux CD reconciles every manifest from git, secrets live encrypted in-repo with SOPS and age, CloudNativePG runs the databases, and Prometheus pages my phone when something breaks. There’s a photo tour.
For the keyword scanners: EKS, GKE, k3s, Flux CD, ArgoCD, Argo Workflows, GitHub Actions, Terraform, SOPS, Traefik, cert-manager, Longhorn, CloudNativePG, Prometheus, Grafana, Loki, Tailscale. Every one of them attached to a cluster above or on the resume.
The day-2 receipts
Carried Cloaked from hundreds of users to hundreds of thousands under SOC 2, ISO 27001, and ISO 27701, without a rebuild and without a second ops hire, by running capacity and cost myself on the EKS platform I built.
Moved Cloaked off its legacy PaaS and made CyberGRX’s risky deploys routine, with users on both products the whole time, by migrating behind the running app and by writing a blue/green operator that kept the old version a selector flip away.
Kept the pager quiet on a homelab that serves real traffic, with Prometheus reaching my phone only for what needs a human, by shipping every change from git, letting certs and backups run themselves, and replicating storage with Longhorn.
“We were deploying C# on vSphere when Jake joined our first Node.js backend team. He had been using Docker, we identified Kubernetes as our next step, and Jake hit the ground running. Our Node services had the fastest release cycle in the company.”
Architect, Food Service Warehouse, 2015
Sound like the role you’re hiring for?
Email me and let’s find out. My resume digs into the details if you need more convincing.
Or browse my projects & ventures. Screening with an AI assistant? Point it at ai.jakegaylor.com.