Project log

How I set up a GKE cluster from scratch in europe-west3 to host shared GitLab CI/CD runners for the organization, including Workload Identity, Node Auto Provisioning, and IAP-based jump host access.

Replaced managed node groups with Karpenter for cost-driven autoscaling. Consolidation mode cut idle compute by ~30%.

Wired External Secrets Operator to GCP Secret Manager using Workload Identity — no static keys, secrets sync on schedule.

Structured Flux tenancy so each team owns their namespace resources while platform team retains cluster-scope control.

Extended Prometheus retention from 2 weeks to 13 months by adding Thanos Sidecar, remote object storage, and a Thanos Query layer.

Wrapped the upstream EKS module with org defaults — VPC layout, IRSA, Karpenter, add-ons — so teams provision clusters via a single module call.

Moved a Kotlin microservices platform from script-driven deployments to ArgoCD, with Datadog for observability.

Jenkins was flaky and nobody trusted it. I was one of the people who made it reliable again — and that foundation made it possible to migrate to GitLab CI without losing ground.

Provisioned Azure infrastructure for a connected vehicle data pipeline — AKS, Event Hubs, and VM-based workers managed via Terraform and Ansible.

Deployed kube-prometheus-stack across AKS clusters, built custom Grafana dashboards for application SLOs, and set up PagerDuty alerting.

Built the CI/CD pipeline for our microservices suite — Jenkins pipelines, Docker image builds, ECR push, and ECS blue/green deploys.

Built and deployed Spring Boot microservices for a retail client on AWS ECS, moving from a monolith to domain-separated services.

Built an internal ops dashboard for a logistics client — real-time fleet status, delivery metrics, and alert management in React backed by a Spring Boot API.