Showcases

Built a self-service platform for 100+ microservices: reusable GitHub Actions workflows, ArgoCD + Argo Events + Argo Workflows as the core. Result: −9% CI time, config drift gone, teams deploying without waiting on infra. Same pattern across 6+ client environments.

Observability stacks (Grafana, Prometheus, Loki, OpenTelemetry) across four enterprises; 4 Golden Signals dashboards; node warm-up from 3 hours to seconds for 1000+ pods. Scale doesn’t have to mean a black box.

I’ve architected greenfield AWS for 350k+ users and 1.2B+ transactions, built self-service CI/CD for ~20 services from day one, and helped product and platform scale together. Platform engineering and developer experience are what make that sustainable.

I’ve run SoW-based AWS assessments, MORA/MODA-driven analysis, and designed both target architecture and migration architecture for multiple clients. Outcome: a clear path from current state to production-ready cloud, not just a pile of recommendations.

KEDA + Karpenter for autoscaling and bin packing; GCP preemptible for dev/stage (−10% compute); cost visibility and right-sizing. We’ve taken infra costs down ~5% where it mattered without sacrificing reliability.

Atlantis (or similar): plan and apply right in the pull request. Reviewers see terraform plan output, approvals gate apply, and state stays consistent. Terraform changes become reviewable and safe instead of “run plan locally and hope.”

Company-wide Terraform registry with versioned modules, plus IaC training for dev and QA. Teams consume infra as code on their own. Same idea with Zero Trust: Cloudflare + IaC + CI/CD so access and boundaries are explicit and auditable.

From inference on Kubernetes to GPU/spot usage and scaling: right-sizing, spot mixes, and observability for AI pipelines. I’ve optimized infra where AI workloads run so cost and performance are predictable instead of a black box.