I've been studying Kubernetes and have local clusters running. I enjoy networking and security, and I'm considering a leaked-password checking service as a project. Would that be useful for a junior job search, or should I focus on something else?
What would make a good Kubernetes project for a junior?
Answers
The password checking service sounds like a solid project, especially if you want to combine security with Kubernetes. Keep the first version small and use a test dataset rather than making the whole thing depend on a production-scale breach database.
Then show how you deploy and secure it: application and database workloads, network policies, certificates, and a way to update the service. Document the privacy design and its limitations too. That gives someone reviewing it more to discuss than just a collection of manifests.
Reply to this answer
Build a deployment pipeline around any modest app. Use CI to test and build it, then use Argo CD and Helm to deploy it into separate development and production-like environments. Terraform could provision the infrastructure if you want to practice that too.
For a portfolio, show how a change moves between environments and what happens when a deployment needs to be rolled back. Two namespaces on a local cluster can be a starting point. Separate clusters add another layer of complexity, but aren't required to demonstrate the workflow.
7v+gaugepocket_483:
What does production-like mean if everything is running locally?
Reply to this answer
Since networking is the fun part for you, make that the centerpiece. Deploy a few services, restrict traffic with network policies, and document how you diagnose a connection that should work but doesn't.
You could compare how service routing and policies behave with your chosen setup, including Cilium if you want to explore it further. A clear troubleshooting walkthrough would show more than a diagram alone.
Reply to this answer
Don't overlook observability just because the cluster is local. Add metrics and logs, then create a controlled problem, like an app returning errors or a workload running out of resources. Use the data to find it and write down what you checked.
OpenTelemetry and a monitoring stack are options, but you don't need every component at once. Pick a question you want to answer first, then add the tools needed to answer it.
Reply to this answer
Another option is to run a few ordinary self-hosted applications and maintain them rather than inventing one big service. Deploy updates, deal with configuration, and document how to recover the setup. The ongoing maintenance can be a useful project in its own right.
Reply to this answer
Add your answer
Answers are reviewed before publication.
6u~ticketlantern_359:
Would you put more time into the security side or the deployment pipeline?
Tarnel~5q:
I'd get a small, working service deployed first, then pick one security feature and explain it well. A basic pipeline is more useful than a half-finished stack of tools.