Blog

The first time my homelab went fully down: a circular bootstrap dependency where the cluster couldn't pull images because its DNS server was a pod inside the cluster that hadn't started yet.

Automating OS updates and kernel reboots across a homelab Kubernetes cluster, and muting the alerts so routine overnight reboots don't wake me up.

Adopting an existing Helm release under ArgoCD on a live Kubernetes cluster, plus the Hubble TLS drift problem that surfaces along the way.

Adding Datadog to a homelab for resume reasons turned into a cascade of platform hygiene work I'd been overlooking

Building an AI-powered runbook generator that also gave me a reason to actually configure Prometheus alerting on the homelab cluster.

Containerizing a Next.js app and setting up a full GitOps CI/CD pipeline with GitHub Actions, GHCR, and ArgoCD, running alongside the existing AWS Amplify deployment.

Setting up ArgoCD for GitOps, migrating from Flannel to Cilium, and wiring up Gateway API with cert-manager and Let's Encrypt for real HTTPS on a homelab Kubernetes cluster.

Deploying the kube-prometheus-stack on a homelab Kubernetes cluster, exposing Grafana remotely over Tailscale, and a swap gotcha that took down the cluster after a reboot.

Setting up KVM virtualization and a three-node Kubernetes cluster on a Dell micro PC: the decisions, the why, and the gotchas.

The technical decisions behind this portfolio (Next.js, MDX, TypeScript, and AWS Amplify) and why each one made sense.