Site Reliability Engineer
I help teams run Kubernetes platforms that are boring in production — predictable releases, alerts that point at the cause, and changes that can be rolled back. Hands-on work, not slide decks.
Multi-hop Kubernetes upgrades, CNI swaps (Calico to Cilium without kube-proxy), ingress-nginx to Gateway API, storage migrations (Longhorn to LINSTOR). Rehearsed first, with a rollback plan and an agreed downtime budget.
SLOs and error budgets, burn-rate alerting, Prometheus, VictoriaMetrics and Grafana. Less alert noise, faster root cause, dashboards on-call engineers actually use.
Calm incident handling, runbooks for the failures you already know, blameless postmortems with action items that get closed.
Argo CD app-of-apps, GitLab CI pipelines, Helm and Kustomize, multi-environment promotion and a safe release path from commit to production.
Trivy and SonarQube in the pipeline, secrets in Vault, Pod Security, least-privilege CI tokens, strict CSP and security headers at the edge.
Terraform and Terragrunt on AWS (EKS, RDS, VPC, IAM), Ansible for bare metal and Proxmox, hybrid setups where part of the stack lives outside Kubernetes.
A read-only review of a cluster or the whole infrastructure — risks, quick wins and a prioritised roadmap, delivered as a written report.
An SRE or DevOps role in a product team, remote from Tbilisi.
A fixed scope with a clear finish line — an upgrade, a migration, an audit or a new pipeline.
Hours for design reviews, a second opinion on a risky change or help during an incident.
30 minutes about the problem, the stack and the constraints.
I look at the real state before proposing anything.
Steps, risks, rollback and the point of no return — agreed before any change.
Changes in agreed windows, with backups and verification after each step.
Runbooks, decision records and documentation, so the team owns the result.
Rates depend on scope and format — tell me about the task and I will reply with an estimate.