Kubernetes: too complex? Materials from DevOps Deflope No. 62 (Category #PlatformEngineering)

I was a guest on DevOps Deflope, together with Alexander Kachmashev from Tochka Bank. In episode 62, released on September 13, 2026, we talked about why launching an application in Kubernetes is easier than living with everything that comes afterward.

We discussed: - Kubernetes complexity. What happens when the familiar Helm chart gives way to networking details and leaky abstractions. - Internal platforms. What work they take off developers' plates, and who takes that work on instead. The number of teams onboarded does not yet tell us how comfortable those teams are with the platform. - AI agents in the cluster. Reading state, offering advice, and changing infrastructure are three different levels of trust. We discussed checks, limited privileges, and changes made through GitOps.

Episode materials: 🎧 Apple Podcasts , Yandex Music, Spotify 📌 The episode page on my site and the text transcript 📖 My preparation longread: where Kubernetes complexity moves

Which parts of that complexity are still handled by developers in your organization, and which have moved to the platform team?

#PlatformEngineering #Kubernetes #DevOps #AI #Engineering