Viktor E.

I keep production Kubernetes healthy — so your team ships instead of firefighting. Day to day I run a production Kubernetes platform: ~12 clusters (prod + staging) across two data centers (US + EU), ~120 nodes, ~2,000 pods, serving 40+ internal engineering teams. I've run production Kubernetes since 2019 — from hand-rolled upgrades with manual certificate regeneration in the early days to today's fully automated, GitOps-driven Cluster API flow. I lead the platform team that owns cluster lifecycle, CI runners, registry, secrets and observability — and I'm hands-on in the incidents that matter. I work best as the ongoing reliability engineer you keep on retainer: the person who owns uptime, responds when something breaks, and removes problems before they page you at 2am. I use AI where it makes operations faster and safer — with checks I design around it — not because it's fashionable. Kubernetes support & reliability: - Ongoing cluster maintenance and fleet-wide upgrades (Cluster API, Cilium, Gateway API) - GitOps everything — platform changes are PR-driven (Argo CD, Helm) - CI/CD on self-hosted GitHub Actions runners (ARC), with cross-site burst capacity - Monitoring and alert hygiene (Prometheus, Thanos, Grafana) — signal, not noise - Incident response and root-cause fixes, not just restarts - Cost-aware, right-sized infrastructure Reliability, from real incidents: - Took a production alert channel from ~20 constantly-firing alerts to zero in one day — ~80% traced to one broken teardown path, which I fixed and turned into a runbook - Caught a runaway cross-site data loop burning ~$360/day and stopped it with one GitOps change within two days - Cut an internal account's cloud run-rate ~25% ($8.4k to $6.3k/mo), and separately ~$18k/year off a client's AWS bill without touching their workloads Why trust me with production, as a freelancer: On Upwork I was the server administrator for a German agency's fleet — 227 hours, 5.0 rating, an ongoing support relationship. I found and cleaned a live intrusion (GSocket implant, PHP backdoors, a backdoored plugin talking to a C2 server) and migrated 25 sites to a new server with zero downtime. I'm also an open-source contributor to Harbor, the CNCF container registry (goharbor/harbor#23972). How I work: - AI agents handle the repetitive 80% — I design the checks that make their output trustworthy - Clarity and maintainability over buzzwords - I explain decisions in plain language - I treat infrastructure as a product, not a side task If you need someone dependable to keep your Kubernetes running — and to pick up the phone when it doesn't — let's talk about an ongoing arrangement.