2023Meridian
Made infrastructure boring on purpose
Infra-as-code, gated CI/CD and an observability stack with SLOs. On-call went quiet; the team shipped on Fridays again.
Ops← All work
99.98%measured uptime
The challenge
Meridian's infra was hand-provisioned, deploys were manual, and on-call was loud. The engineering team had stopped shipping on Fridays — and morale was following suit.
Our approach
How we got there.
01
Everything as code
Terraform modules for every environment, with drift detection and policy-as-code gates. No more snowflake servers.
02
Gated CI/CD
Automated test, security scan and preview deploy on every PR. Production deploys became a button press with automatic rollback.
03
SLOs that mean something
Observability stack with error budgets, burn-rate alerts and runbooks. On-call went from reactive firefighting to quiet weekends.
Outcomes
What changed.
99.98%measured uptime
−80%on-call pages per month
Frideploy day restored
Stack
Tools and technologies used across this engagement.
- Terraform
- AWS
- Datadog
- ArgoCD
- Pulumi
- PagerDuty