Case study · Healthcare payments · US · AWS · Delivered
A payments worker that scales with its queue
A US payments company serving dental practices. The client is described by sector and region only.
The challenge
- Payment jobs had to survive restarts with nothing lost
- Uneven load: a fixed fleet wastes money or falls behind
- The pipeline had to reach AWS without stored keys
What we delivered
- Queue worker on Fargate
- Deletes a job only on success; drains in-flight work on deploy
- Failure handling
- Four retries, then a dead-letter queue with an instant alarm
- Backlog auto-scaling
- 1–4 workers in production; QA and UAT scale to zero when idle
- Keyless CI/CD
- Terraform for three environments; GitLab CI with OIDC into AWS
Outcome
The worker absorbs spikes on its own, test environments cost nothing when idle, and no cloud keys are stored in the pipeline.
Technology
AWS ECS Fargate · SQS · ECR · CloudWatch · Terraform · GitLab CI (OIDC)
Services used
- DevSecOps · Keyless deploys and pipeline checks
- Terraform & Infrastructure as Code · Every environment in code
- Cloud Cost Optimization · Read-only review, savings you can verify
