Solutions

Real solutions

Infrastructure guides from real-world projects.

Deep dives into DevOps challenges, infrastructure patterns, and proven solutions. Learn from work that's already live in production.

Complete SRE Stack: Monitoring, Security & Disaster Recovery

Your app goes down on Friday night. You don’t find out until Monday when a customer complains.

This doesn’t have to happen. The same SRE principles that keep large systems alive — observability, threat detection, and disaster recovery — work at any scale. They’re the same tools, just sized to your infrastructure.

This is the complete SRE stack: observe your cluster, catch security threats before they spread, and restore from backups in minutes. All open source. Deployable on any Kubernetes cluster.

Read solution

Implementing Secrets & Zero-Trust in Kubernetes

Your database password is in a ConfigMap. Your API key is in a git repo. Your cloud credentials are in plaintext environment variables.

This is how most Kubernetes clusters look in the wild. And it’s how breaches happen.

Zero-trust means: don’t trust anything by default. Encrypt secrets at rest. Network-isolate workloads. Control access explicitly. Only then grant it.

This article covers a complete secrets and zero-trust strategy: encryption, network policies, access control, credential rotation, and scanning for leaks. We’ll use real examples from a production-grade setup.

Read solution

High Availability Kubernetes on Ubuntu for GPU-Accelerated AI Workloads

Built a 6-node HA Kubernetes cluster with 3 control planes + 2 QEMU VMs on Ubuntu 22.04 LTS. Covers cluster architecture, GPU driver support, operational flexibility, and production-ready setup for AI workloads.

Read solution

Production Kubernetes Networking and GitOps: Cilium, Flux CD, and Democratic CSI

Migrating a bare metal Kubernetes cluster from Flannel+MetalLB to Cilium eBPF networking, implementing Flux CD GitOps with SOPS+age secret encryption, and provisioning persistent storage via Democratic CSI and TrueNAS iSCSI.

Read solution

Private AI Platform: LiteLLM, Open-WebUI, Qdrant, and RAG on Kubernetes

A self-hosted AI stack on Kubernetes: LiteLLM API gateway, Ollama GPU inference on a GTX 1070, Open-WebUI chat interface, Qdrant vector database, and a RAG pipeline that queries 239 chunks from 98 knowledge base files — plus a Python CLI and MCP server wired into Claude Code.

Read solution

Kubernetes AI Platform: GitOps Migration

Why We Merged Our Kubernetes Repositories (And How)

Read solution

Building an End-to-End GitOps CI/CD Pipeline: From Docker to Flux

Eliminated manual deployments entirely — zero-touch pipeline from Docker build through GitHub Actions CI to Flux CD auto-sync on Kubernetes.

Read solution

Building a Portable DevContainer Environment: A Complete Journey

Reduced dev environment setup from hours to minutes — fully reproducible DevContainer with Mise, Chezmoi, and Neovim that spins up identically on any machine.

Read solution

Building a Secure Arch Linux Server with Full Disk Encryption

Hardened a bare-metal server to production security standards — LUKS2 full-disk encryption, pure Wayland, zero unnecessary services, Tailscale for remote access.

Read solution

Securing LAN Networks with pfSense and Mullvad VPN

937 Mbps throughput maintained through full VPN tunnel — network-wide traffic encryption with pfSense and Mullvad, no per-device clients, enterprise firewall rules.

Read solution