Writing

A walkthrough of moving from a slow Jenkins-based pipeline to ArgoCD with blue-green deployments on EKS — and what surprised me along the way.

The patterns I use to write Terraform modules that are flexible enough for multiple teams but simple enough that nobody avoids them.

Loki is powerful but easy to misconfigure. Here is the setup that finally worked for us across 20+ services without blowing the storage budget.

Latency, data residency, Route53 quirks, and RDS cross-region replication gotchas from our migration to a three-region active-active setup.

Practical conventions for shell scripting in a DevOps context: error handling, logging, idempotency, and how to avoid the traps that bite you at 2am.