I help small engineering teams run production infrastructure — AWS, Kubernetes, CI/CD and observability — without hiring a full-time DevOps engineer.

Book a call

What I fix

Deploys are slow and nobody's sure why.

I map the pipeline end to end, cut the waste, and document what's left so the next person doesn't have to re-derive it.

You find out about incidents from your customers.

Metrics, logs and traces with Prometheus, Grafana and OpenTelemetry — and alerts that fire on symptoms your users feel, not on CPU.

Terraform no longer matches what's actually running.

I reconcile the drift, import what was created by hand, and get you back to infrastructure you can change without holding your breath.

Your Kubernetes bill grows faster than your traffic.

I check requests, limits and autoscaling against what the workloads actually use, and hand back the changes ranked by what they save.

Evidence

Merged upstream

Pull requests merged into Apollo Router and Apollo Federation — the routing layer behind federated GraphQL in production, written in Rust — and the NestJS GraphQL integration.

every pull request on github →

Operated in production

  • A client system I have run in production since 2023 — around 500 requests per second on average, peaking above 2,000.
  • Proxmox and Oracle Cloud clusters of my own, each with separate development and production environments.
  • Kubernetes on EKS, GKE and OKE; pipelines with GitHub Actions and ArgoCD.
  • Day job: production troubleshooting for teams running Apollo Router and Apollo Federation.

Certified

How it starts

Infrastructure assessment

$500 · fixed price · two weeks

I review your cloud setup, pipelines and monitoring, then deliver a written assessment: what is there now, what is most likely to break, and what to fix in what order. You keep the document whether or not we work together.

Read-only access is enough. No changes to your systems.

What comes after

Larger engagements — around 80 hours — start at $4,000. Prices in USD; EUR on request.

Availability

I take on one or two engagements at a time, around 10 hours a week each, working asynchronously. Available for calls 10:00–16:00 UTC, Monday to Friday — afternoons in Central Europe, mornings on the US East Coast.

I don't offer on-call or incident response.

If that fits, book a call and we'll scope the assessment.

Book a call