Rebuilding Cloud Infrastructure at Advocara
From fragile deploys to 99.95% uptime and 85% faster releases
Outcome
85% faster deployments · 99.95%+ uptime · $1,800/yr saved
The Problem
Advocara's deployment pipeline was manual, slow, and error-prone. Infrastructure was ad-hoc, with no observability, inconsistent environments, and security gaps in the auth layer.
The Approach
- 1
Audited existing AWS setup — identified over-provisioned resources and missing automation.
- 2
Migrated to ARM-based EC2 (t4g.small) + SSM Session Manager, eliminating bastion host costs.
- 3
Built Terraform modules for EC2, ALB, API Gateway, RDS, and ECS Fargate — fully reproducible infra.
- 4
Wired GitHub Actions CI/CD with automated testing, staging deploys, and production gating.
- 5
Set up Grafana dashboards using CloudWatch, Prometheus, and Loki for real-time incident alerting.
- 6
Rebuilt the auth layer with NextAuth.js + Cognito — blocked 75% more unauthorised requests.
- 7
Introduced a unified DatabaseManager to eliminate N+1 patterns and centralise CRUD logic.
The Result
Deployment performance improved 80–90%. Release velocity up 40%. Incident resolution time down 25%. System maintained 99.95%+ availability post-launch.