Many people on X are going to get offended by this 😅 These days I keep seeing posts like: “AI can write Terraform code.” “AI can write Kubernetes manifests.” “AI can write Dockerfiles.” “AI can automate coding.” And then people say: “Now you need to learn deep networking.” “Troubleshooting is more important than ever.” “Just use AI, you don’t need to know everything.” My friend… start somewhere. If you’re new to DevOps or any technical field, PLEASE learn your basics. Don’t directly jump into Kubernetes, copy a manifest from Claude, apply it and think you know Kubernetes. Eventually, you’ll get fucked when something breaks. Use AI as a companion, not as a copy-pasting machine. If AI generates a Kubernetes manifest, understand what it generated. If AI writes Terraform, understand what resources it is creating and why. If AI gives you a Dockerfile, understand what is happening inside the image. Learn the important tools. Learn networking. Learn Linux. Learn troubleshooting. Learn how systems actually work under the hood. And yes, use AI. Of course. The interesting part is that I still see people saying: “AI will do everything. You don’t need to learn these things anymore.” Really? Then who is going to optimise that Terraform code according to your actual context? Who is going to debug the Dockerfile when the build suddenly breaks? Who is going to understand why your Kubernetes workload keeps crashing? Who is going to make the decision when you get paged at 3 AM? Who is going to decide whether the AI-generated solution is actually safe to deploy? If AI can do all of this without technical understanding, why don’t companies just hire a 10–15 year old kid, give them Claude Max and let them run production? Because that’s not how engineering works. AI can generate code. But understanding the problem, providing the right context, validating the output, debugging failures and making the final decision still requires engineering knowledge. And honestly, I don’t think people should blindly believe everything they see on X. Not even me. Keep learning. Use AI. Go deeper as your experience grows. But don’t skip the fundamentals just because AI can generate the syntax for you. You still need to understand what you’re building.

It's been one hell of a week in my local homelab where I experiment, break things, troubleshoot, and learn about Kubernetes, cloud, and DevOps. Over the last few days I completed: ✅ ArgoCD with Image Updater ✅ Cosign keyless signing ✅ Dynamic GitHub Actions matrix ✅ Prometheus + Grafana monitoring ✅ Centralized logging using the EFK stack One thing I realized is that building something from scratch teaches you way more than following a tutorial. Some things I learned that weren't covered in most beginner tutorials: Using the App of Apps pattern in ArgoCD Managing deployments with Helm charts (both my applications and the EFK stack) Understanding how all the components actually work together instead of just deploying manifests I didn't end up using Fluent Bit Lua scripts because my use case didn't require them. Most of my time wasn't spent writing YAML—it was spent troubleshooting. Debugging why something doesn't work teaches far more than watching another "Deploy EFK in 10 minutes" video. The homelab is slowly starting to look like a production-grade platform, and there's still a lot more to build. 🚀