Nokia
Working Student DevOps Engineer · Jul 2024 – Present · Poland
AWS infrastructure automation — VPC, EKS, RDS, Crossplane, Terraform
- Problem
- Monolithic Terraform applied VPC, EKS, RDS, and product Helm in one run — coupled infra and app deploys, external EKS extensions instead of built-in addons, no continuous reconciliation.
- What I did
- Built a Terraform-provisioned control-plane EKS (built-in AWS addons, Crossplane pre-installed) with compositions and XRDs for prod/dev envs; separated product deployment from infrastructure.
- Result
- Control-plane cluster in ~15 min; prod + 2 dev infra envs deployed via compositions/XRs with GitOps reconciliation; standalone AWS resources integrated through XR parameters.
GitOps for platform and application delivery
- Problem
- Infrastructure and application changes lacked auditable, continuously reconciled delivery from Git.
- What I did
- Implemented GitOps workflows with Crossplane compositions and XRDs for platform and application delivery from the control-plane cluster.
- Result
- Auditable, Git-driven infrastructure and application delivery with continuous reconciliation.
Infrastructure monitoring and alerting — Prometheus, Grafana
- Problem
- Datadog was the sole monitoring stack for EKS; the team needed a self-hosted alternative after moving away from the licensed solution.
- What I did
- Deployed kube-prometheus-operator with ALB ingress, EBS-backed storage, product ServiceMonitors, team Grafana dashboards, and CloudWatch integration via IRSA.
- Result
- Production-ready monitoring on every Helm deploy with no extra configuration — setup time reduced several times over.
AWS cost optimization across accounts
- Problem
- Nine separate ALBs for cluster ingresses; dev cluster nodes running 24/7 outside business hours.
- What I did
- Scheduled node autoscaling (scale to 0 off-hours, autoscale during work hours) and consolidated all ingress target groups onto a single shared ALB.
- Result
- Dev compute scaled down automatically outside business hours, substantially reducing infrastructure costs; ALB costs cut ~90% (9 → 1).
Jenkins CI/CD pipeline maintenance
- Problem
- Existing Jenkins pipelines and product Helm charts needed ongoing updates as the product and deployment workflows evolved.
- What I did
- Maintained Jenkins configurations, updated deployment pipelines, and integrated product changes into Helm releases.
- Result
- Reliable product deployments kept in sync with infrastructure and application changes across environments.