I avoided learning Kubernetes for two years.

"I'm an ML engineer, not a DevOps person," I told myself.

Then I needed to deploy a model that scaled from 100 to 10,000 requests per hour depending on the time of day. And I realized I was the bottleneck on my own team.

Here's the reality: "Must know Kubernetes" is in almost every ML engineer job description now. Not deep K8s expertise — nobody expects you to configure networking from scratch. But deploying a model on K8s, scaling inference pods, managing GPU resources, and understanding KubeFlow or KServe? That's expected.

The combination of ML + DevOps is the most in-demand skill mix I've seen. Because teams are tired of having brilliant model builders who can't get anything into production without hand-holding.

You don't need to become a K8s expert. But if you can deploy your own models, scale them, and monitor them? You just became worth 30% more.