Kubernetes policies are usually discussed in the context of cloud workload, but manufacturing environments introduce a different kind of risk. Here, configuration mistakes can have physical consequences, not just failed deployments.

In this talk, we explore how policy-as-code can be extended beyond data centers into a manufacturing environment. Using Kyverno as a policy decision engine, we demonstrate an architecture where policies govern electrical safety thresholds at production plant

Sensor readings such as voltage, power consumption, and temperature are published into Kubernetes-native resources. Kyverno evaluates these signals to enforce safety policies, trigger alerts, and emit escalation signals when critical thresholds are crossed. At the edge, lightweight Kubernetes clusters running on Raspberry Pi devices act as local policy decision points.

This session shares practical design choices and lessons learned from applying Kubernetes governance to industrial and IoT workloads.