· Cloud
Healthcare ISV
Problem
Regulatory requirements mandated on-prem data residency but the dev team needed cloud-like velocity.
Outcome
Private Kubernetes platform delivered; developer cycle time reduced from 3 days to 4 hours.
Key metrics
- Dev cycle time
- 4 hrs
- Previous cycle time
- 3 days
- Cluster availability
- 99.9%
What we did
The client processes sensitive health records subject to strict data residency rules that ruled out public cloud for production workloads. Their engineering team was deploying by SSH, with no staging environment and no consistent tooling between developers.
We designed a bare-metal Kubernetes platform hosted in their own data centre: three control-plane nodes for HA, worker pools separated by workload class, and a GitOps delivery model using ArgoCD so every environment state lived in version control.
The developer experience layer mattered as much as the infrastructure. We introduced a standardised Helm chart library for their services, a local dev environment that mirrored production using kind, and a short feedback loop where a pull request triggered a preview environment automatically.
After onboarding the first three services, the team's median cycle time — from code merge to running in staging — dropped from three days to four hours. Six months in, the platform was running twelve production services with no unplanned outages.
Have a similar problem?
Start a project →