I build the platform behind the product.

I'm a Cloud Engineer who works on the systems that keep multi-region Kubernetes platforms running. My work covers AWS infrastructure, Kubernetes upgrades, automation and operational tooling, with hands-on experience across five production clusters.

An iceberg. The small tip above the waterline is the product, which is all a customer sees. The much larger mass under the water is the platform, layer by layer: Kubernetes (Amazon EKS, Gateway API, Velero), GitOps (Argo CD, Helm), CI/CD (GitLab CI, BuildKit, Python), observability (Datadog, Opsgenie, Checkly), secrets (External Secrets, 1Password), and cloud infrastructure underneath (AWS, Cloudflare, Terraform).An iceberg. The small tip above the waterline is the product, which is all a customer sees. The much larger mass under the water is the platform, layer by layer: Kubernetes (Amazon EKS, Gateway API, Velero), GitOps (Argo CD, Helm), CI/CD (GitLab CI, BuildKit, Python), observability (Datadog, Opsgenie, Checkly), secrets (External Secrets, 1Password), and cloud infrastructure underneath (AWS, Cloudflare, Terraform).

Projects

  1. Onboarding customer files to Cloudflare R2

    One pipeline moves a customer's files from the server's disk into its own R2 bucket.Cloudflare R2 · Helm · Terraform · Datadog
  2. Shared ingress on Envoy Gateway

    One Envoy Gateway per cluster routes every customer by server name.Envoy Gateway · Gateway API · AWS Network Load Balancer · Helm
  3. CLI tool to automate server lifecycle

    A CRM event opens the merge request that creates, updates or deletes a server.Python · uv · GitLab CI · Argo CD
  4. Password rotation without a restart

    The support password now goes from 1Password straight into running servers.External Secrets Operator · 1Password · Helm · AWS Secrets Manager
  5. Moving servers between AWS accounts and regions

    Customer servers moved between AWS accounts and regions in one window.Velero · Amazon EBS · AWS KMS · Kubernetes
  6. Less alert noise, lower cloud cost

    236 alert rules became 46, and one cluster ran on one node fewer.Datadog · Terraform · Kubernetes · Amazon EKS