Skip to content

03 · Service

Backend, Cloud & Platform Engineering

The systems underneath: event-driven, fault-tolerant, and built for compliance-grade workloads.

The approach

When money or health data moves through a system, correctness stops being a preference. This is the half of my work that lives underneath the product: queues, reconciliation, audit trails, and the infrastructure they run on.

On the fintech side that has meant ACH and wire pipelines on pacs.008, camt.053, and camt.056, SFTP file processing that replaced manual banking workflows, and event-driven queues for operations that cannot be retried carelessly.

On the platform side it has meant Terraform environments standing up VPCs, RDS, ECS Fargate services and Lambda authorizers from scratch; GitHub Actions CI/CD with OIDC keyless auth and fail-closed guards against misrouted deploys; and a HIPAA Safe Harbor de-identification pipeline that strips PII while keeping aggregate analytics intact.

What you get

  • Service boundaries drawn where the domain actually splits, not where the org chart does
  • Queue and event architecture — BullMQ, RabbitMQ, EventBridge — with retry, dead-lettering, and idempotency
  • Infrastructure as code in Terraform, reproducible across dev, staging, and production accounts
  • CI/CD on GitHub Actions with keyless OIDC auth, branch-to-environment routing, and automatic rollback
  • Compliance-grade data handling: audit trails, reconciliation, and de-identification pipelines
  • Cost work — Fargate Spot, right-sized RDS, and the monitoring to know when it drifts

Where this shipped

Other services