Software Engineer · DevOps · 4+ yrs
I build software and run it in production — the app itself, and the CI/CD pipelines, Terraform and Kubernetes underneath it on AWS and Azure. I use LLMs as a power tool inside the pipeline, not as a crutch. Everything below is real — in production or open source, no demos.
A micro-frontend banking dashboard taken from source to a Kubernetes cluster on Azure — twelve resources in Terraform, and a Jenkins pipeline per repository that tests, builds and rolls out on every commit, failing the build when a rollout doesn't converge. AKS pulls images through Managed Identity instead of credentials in a secret. Six repos; infra torn down after delivery.
Vehicle-restriction checker for Colombia — one tap to know if you can drive today, across dozens of cities. Built SEO-first so it ranks: ~50K+ monthly visitors, 90+ Lighthouse.
A conformance gate for CI/CD, packaged as a GitHub Action. It validates microservices and decides on its own whether to block a merge — deterministic rules combined with LLMs, so quality and traceability are enforced before code lands. Open source — read it.
A segmented AWS network as reusable Terraform modules — public, private and data tiers across two AZs. The data tier can't reach the internet by absence, not by a deny rule: its route table has no default route. Managed without SSH or a bastion via SSM endpoints, and covered by terraform test.
Terraform for a static site with its own domain and HTTPS — S3, CloudFront, ACM and Route 53 in one apply. The bucket stays private and only answers signed requests from the CDN, and GitHub Actions ships every push through OIDC, so no AWS keys are stored anywhere. Running live at xyzr.online.
A fitness app built for body-recomposition: planning and tracking that stays out of your way. Designed and shipped end-to-end, from UI to deploy.