Bank Statement Service — Scalable Platform

The project

A bank statement processing service running across production, UAT and dev EKS clusters, backed by RDS Postgres, Amazon MQ and S3, with an on-prem VPN link. The platform processes statement workloads through a fleet of browser agent workers, so scaling behaviour and release safety directly affect throughput and uptime.

What I did

• Migrated stateful browser-agent workloads from StatefulSets to Deployments so pods could scale in and out cleanly.
• Wired KEDA event-driven autoscaling to queue depth and consumer lag, so the worker fleet follows real demand instead of a fixed replica count.
• Fixed a CI pipeline image-tagging bug that was silently shipping stale images to production.
• Managed Helm charts and Bitbucket Pipelines for repeatable releases across the three clusters.
• Led incident response on a multi-day production outage — diagnosis, recovery and follow-up hardening.
• Authored a blue-green deployment feasibility report as the path to zero-downtime releases.