Cloud / platform · Veganext · May 2026
Multi-region Azure platform for document processing
Active-passive, multi-region Azure deployment of an intelligent document processing platform replacing a legacy OCR product.
At a glance
- Complete deployment written as repeatable infrastructure-as-code, runnable end to end in 30-45 minutes: network, secrets, data, compute, validation.
- Two Kubernetes clusters - autoscaling primary, warm standby secondary - with a private-only data plane and no secrets in code or container images, using workload identity federation instead.
- Documented cost/risk trade-off on the messaging tier that saved around $1,330/month on one environment, with the manual failover procedure written up as a runbook.
Stack
Azure (AKS, Key Vault, Front Door + WAF, Service Bus, PostgreSQL, ACR, Log Analytics)Bicep / ARMPowerShellKubernetesHelmDocker
Repo and demo
Client work, code is private. The architecture and the evaluation method are covered above, and I'm happy to walk through the code in an interview.