Cloud & DevOps Engineer
LinkedIn | GitHub
PR #2365 (merged) – Backward-compatible migration for renamed OTLP exporters (otlp → otlp_grpc, otlphttp → otlp_http)
- Rewrote exporter definitions and pipeline references while preserving named instances
- Added deprecation warnings, Helm unit and integration test coverage, and upgrade documentation
- Resolved merge conflicts and fixed a failing collector test on a local kind cluster
open-telemetry/opentelemetry-helm-charts#2365
PR #2403 (merged) – Decoupled namespaceSelector and objectSelector settings for Pods, Instrumentation, and OpenTelemetryCollector admission webhooks, with per-webhook selector values taking precedence over common selectors.
- Added backward-compatible common and per-webhook selector handling
- Validated with
helm lint,helm templateand chart testing
open-telemetry/opentelemetry-helm-charts#2403
PR #146 (merged) – Added values.yaml validation before running documentation and schema generators
- Added regression tests for both
--fileand--chartmodes - Incorporated maintainer feedback by bumping both generators to 0.3.0 and updating their changelogs
grafana/helm-chart-toolbox#146
From PaaS → Serverless → Infrastructure as Code → Kubernetes-native delivery
2025–2026 | Cloud portfolio evolution
| Version | Architecture | Technologies | Key Achievements |
|---|---|---|---|
| v1 | Django Monolith | Elastic Beanstalk, GitHub Actions | Selected for further development by NIÜ/HSUP; automated CI/CD |
| v2 | Cloud-agnostic Serverless Core | Lambda, Docker, Hexagonal | Documented median latency reduction (~120 ms → 44 ms); ZAP scan with 0 alerts in the recorded scenario |
| v2.1 | Terraform IaC Layer | Terraform, GitHub OIDC | Repeatable AWS provisioning and automated application deployment |
| v3 | Kubernetes-native Runtime | FastAPI, Docker, kind, Helm, Kustomize | kind-based Kubernetes baseline reproduced on AWS EC2, Ingress, HPA, NetworkPolicy, runtime hardening, CI |
The portfolio shows a deliberate progression: starting from traditional PaaS, advancing to cloud-agnostic serverless, adding Terraform-managed infrastructure, and extending into container orchestration with Kubernetes.
Next milestone (v3.1): Terraform-managed AWS EKS, ECR, GitHub OIDC and Helm-based deployment path.
github.com/szelese/v3-k8s-core
A Kubernetes-native delivery project that extends my earlier serverless portfolio into container orchestration.
Key features:
- Isolated Python business core with FastAPI HTTP adapter
- Non-root Docker runtime with Kubernetes security hardening (read-only filesystem, dropped capabilities)
- Local Kubernetes with
kind, Kustomize and reusable Helm chart - Ingress, Horizontal Pod Autoscaler, NetworkPolicy
- Prometheus-compatible
/metricsendpoint + structured JSON logging - Full GitHub Actions CI (pytest, Trivy scan, smoke tests, Helm validation)
This project is intentionally presented as a local Kubernetes runtime portability baseline, reproduced locally and on a fresh AWS EC2 host using kind, not as a production Kubernetes platform.
github.com/szelese/v2.1-agnostic-lambda-core-terraform
A Terraform-based infrastructure automation layer around the v2 serverless core.
Key features:
- Repeatable AWS infrastructure provisioning with Terraform
- Terraform-managed IAM, ECR, Lambda, CloudWatch alarms and SNS notifications
- GitHub Actions deployment workflow
- GitHub OIDC authentication
- Least-privilege IAM design
- Automatic GitHub Secrets management
- Smart smoke tests after deployment
- Initial infrastructure provisioning through
terraform apply; subsequent application deployments are automated
github.com/szelese/v2-agnostic-lambda-core
A cloud-agnostic Python application core adapted to AWS Lambda container runtime.
Key features:
- Hexagonal / clean architecture style separation
- Environment-agnostic business logic
- Dockerized Lambda runtime
- Documented median latency reduction from approximately 120 ms to 44 ms under the recorded Locust test scenario
- OWASP ZAP scan with zero alerts in the recorded test scenario
- GitHub Actions quality gate and version-aware post-deployment smoke test
github.com/szelese/ci-cd-gha-aws
The original monolithic Django project and deployment baseline.
Key features:
- GitHub Actions CI/CD pipeline
- AWS Elastic Beanstalk deployment
- OIDC-based AWS authentication
- Automated deployment workflow
- Post-deploy hooks for database migration and static file collection
- Selected for further development by the Hungarian National Innovation Agency / HSUP
Cloud & Infrastructure
AWS (Lambda, ECR, IAM, CloudWatch, Elastic Beanstalk) • Terraform IaC
Kubernetes & Containers
Docker • FastAPI • kind • Kustomize • Helm • Ingress • HPA • NetworkPolicy • Security contexts
CI/CD & Automation
GitHub Actions + OIDC • Trivy scanning • Smoke testing • Helm validation
Security & Observability
Least-privilege • Non-root containers • Prometheus-compatible metrics • Structured logging
Architecture
Hexagonal design • Cloud-agnostic core
- Certified Kubernetes Application Developer — CKAD
- AWS Certified Developer – Associate
- HashiCorp Certified: Terraform Associate
- AWS Certified Cloud Practitioner
Open to Cloud, DevOps, Platform Engineering and Infrastructure as Code roles in Hungary, the EU and remote international teams.
📩 ervin.wallin at gmail dot com | Let's build reliable, secure and reproducible cloud systems together!
20 years in high-stakes logistics → reliability-first approach to cloud infrastructure.
© 2026 Ervin Wallin