A war story about Terraform state disagreeing with the AWS console — the import attempt that didn't fit, the temporary filter that kept deploys moving, and the pre-prod cleanup that finally let Terraform own the lambda.

Patterns I use to build modular, maintainable GitLab pipelines — with real examples covering multi-environment deployments, change detection, security scanning, container builds, and automated promotion.

Treat your monitoring config like code. CloudWatch alarms as Terraform, metric math for derived rates, per-environment retention tiers, and SNS routing patterns from a real production AWS environment.

Preview environments are a frontend cliché — doing them on real GCP infrastructure is a different beast. The pattern, the costs, and what we'd do differently.