← All case studies
Fintech & paymentsKubernetes platform engineeringDevtron adoption

Kubernetes maturity in three weeks, on a platform clearing 400M+ transactions a month.

BharatPe gives last-mile retailers a single interoperable QR code with zero-fee payments. Their path to production was a script-heavy Jenkins setup that broke whenever the platform scaled, running against self-managed clusters that had quietly drifted apart from one another. They chose Devtron as the delivery platform; we ran the adoption around it.

Client
BharatPe
Industry
Fintech · digital payments
Focus
Devtron adoption on Kubernetes
Time to K8s maturity
3 weeks

The challenge

BharatPe needed to containerise an existing estate and move it onto Kubernetes without denting availability for merchants taking payments. The pipeline standing in the way was held together with scripts, and the clusters underneath it had no shared security posture.

  • Script-heavy Jenkins pipelines that were error-prone and broke during scaling operations
  • Self-managed clusters snowflaking into configuration drift, with uniformity and security concerns across them
  • A learning curve on the new tooling that was pushing out go-to-market dates

What we did

We ran the adoption in four stages, with Devtron as the delivery platform underneath: migrate the existing applications onto it, build templated workflows that clone in one click so a new service starts from a known-good pipeline, enforce security policy centrally rather than per-team, then containerise and roll out to the clusters at scale.

“Devtron helped us adopt Kubernetes quickly, and our dev teams are now self-sufficient, letting DevOps focus on bigger challenges.”— Ravi Ranjan Kumar, DevOps Lead, BharatPe

Most of that was consulting rather than installation: deciding which applications moved first, writing templates the teams would actually reuse, mapping role-based access control onto their existing SSO so permissions followed identities people already had, and working alongside the development teams until they were running their own deploys. The platform pieces doing the heavy lifting were reusable CI/CD templates, one-click rollbacks, drift detection to stop clusters diverging again, and policy-as-code for the security rules.

The result

MTTR dropped from days to under an hour, and lead time for changes went from days to 90 minutes. Release velocity rose 12x, with at least one deployment a day. The platform now handles more than 400 million transactions a month, and the development teams run their own deploys instead of queueing behind DevOps.

Have a similar problem?

Tell us where things stand today and an engineer will reply directly.

Talk to an engineer