I'm learning Kubernetes for work and currently use Colima with kubectl. I've worked with Docker before and understand the basic container concepts, but I'm still new to Kubernetes. I started a hands-on beginner course, but I'm wondering whether it's still relevant in 2026, especially since it has me writing YAML manifests manually. With modern editors and AI tools able to generate configuration, do I still need to learn to write YAML myself? I have about a month to become comfortable enough to collaborate with our platform and AWS teams. What fundamentals, projects, courses, or roadmap would you recommend?
4 Answers
Learn by deploying one small application and deliberately breaking it. Start with a single-node local cluster, deploy the app, expose it with a Service, add configuration and health probes, then troubleshoot failures. Try an incorrect readiness path, very low memory limits, a bad image tag, and a configuration change that requires a restart. Practice using kubectl get, describe, logs, events, and checking endpoints. You don’t need to memorize YAML syntax, but you should become comfortable reading and editing it because real debugging depends on understanding what the manifests actually request.
AI is useful as a tutor and for generating repetitive manifests, but don’t let it replace the fundamentals. You need to understand Pods, Deployments, Services, Ingress, ConfigMaps, Secrets, storage, probes, resource limits, and basic networking well enough to read the YAML and recognize when a generated answer is wrong. The beginner course is still relevant because those core concepts haven’t changed. Use the official Kubernetes documentation alongside it and ask AI to explain concepts, quiz you, and provide troubleshooting scenarios rather than simply completing every task for you.
A good rule is to interrogate the AI instead of just instructing it. Ask why a field is needed, what alternatives exist, and what symptoms you would see if it were misconfigured.
For a one-month goal, focus on practical collaboration rather than trying to master all of Kubernetes. A reasonable progression is: Pods and containers; Deployments and rollouts; Services and DNS; Ingress; ConfigMaps and Secrets; probes and resource requests; volumes and StatefulSets; namespaces and RBAC; then basic observability and troubleshooting. After that, look at how your company uses Helm, infrastructure-as-code, and its deployment pipelines. Learning the conventions in your organization’s repositories will likely be more useful than taking several additional broad courses.
Once you understand the basics, build a slightly more realistic lab with multiple nodes or a managed cluster if you have access to one. Explore how scheduling, networking, persistent storage, and cloud integrations differ from a local environment. You can use AI to create exercises tailored to your setup, but keep it in an advisory role: review its commands, ask for explanations and tradeoffs, and make changes through version control. A deeper exercise such as building Kubernetes components manually can be valuable later, but it is probably not the best use of your first month if your immediate goal is working with a platform team.
The important part is seeing how Kubernetes fails: a pod can be Running but receive no traffic, or be healthy-looking while being throttled or repeatedly OOM-killed. Those scenarios teach more than copying successful examples.