/ selected work
Case studies
Each one follows the same arc — the problem, the constraints I worked under, the architecture I chose, the trade-offs I accepted, and what actually changed. No polish over substance.
Right-sizing and consolidating GCP workloads — moving low-traffic services off managed Kubernetes onto VMs behind Caddy/Traefik and Cloudflare — for a 58% infrastructure cost reduction.
Replacing copy-pasted pipelines with versioned, reusable GitLab CI templates — parallelized and layer-cached — cutting pipeline time ~60% (15 → 4 min) and standardizing builds across repos.
Moving secrets out of GitLab repo settings into HashiCorp Vault (KV-v2), then to OpenBao after the licensing change — keeping the API contract stable so apps and pipelines needed no rewrite.
Standing up Prometheus, Grafana, Loki, and Sentry on a single GCP VM to turn scattered, siloed telemetry into one place to see metrics, logs, and errors — with tracing (OpenTelemetry/Jaeger) as the next step.