___ _
/ _ \ ___ _ __ __ _ _ _ (_) _ __
/ /_)/ / _ \| '_ \ / _` || | | || || '_ \
/ ___/ | __/| | | || (_| || |_| || || | | |
\/ \___||_| |_| \__, | \__,_||_||_| |_|
|___/
Deepening expertise in GitOps workflows, exploring service mesh architectures, and building more resilient disaster recovery patterns. Always asking "why" before implementing "how".
Orchestrated complete ISP migration and network redesign across 17 sites, transitioning from /24 to /28 IP addressing scheme while maintaining zero downtime during business hours. This wasn't just a migration - it was a complete reimagining of our network topology.
Reduced deployment time for network/server policies by 90% through Ansible automation and CI/CD principles. What used to take hours of manual configuration now happens in minutes with full rollback capabilities.
3-node bare-metal Kubernetes cluster running Talos Linux with GitOps workflows via ArgoCD. This is my playground for testing production patterns before deployment. Currently hosting 20+ services including Immich, n8n automation, and local LLM inference.
Implemented ArgoCD across the homelab cluster. The good: declarative configs are magical when they work. The bad: when your git repo becomes source of truth, a bad commit means rapid automated failure. Learned to always test in dev namespace first. Always.
Key takeaway: git revert is your friend, but proper branch protection is your best friend.
You can read all the documentation you want, but nothing beats breaking your own network at 2 AM. Currently running Proxmox, Talos, and enough VLANs to make my switch question its life choices. This week's lesson: MTU mismatches are sneaky.
Deployed Ollama and Llama.cpp with local LLMs integrated into n8n workflows. Now have AI-assisted log analysis and automated incident summaries. CPU inference is painful, but privacy and data sovereignty make it worthwhile. Next step: GPU node.
"Know the why before the how" isn't just a catchphrase and it has saved me countless hours of fixing the wrong problem. Every system has a story, every architecture decision has context. Take the time to understand both.
Automate everything you do more than twice, but understand what you're automating first. The best automation comes from deep understanding of manual processes. Also, always include a manual override. Future you will thank present you.
The homelab isn't just about playing with tech. Your homelab is about having a safe space to fail spectacularly. Every outage is a lesson, every recovery is growth. Document everything, especially the failures.
Infrastructure should be transparent, repairable, and under your control. Whether it's the right to repair hardware or the ability to audit code, ownership means actual ownership. This extends to data sovereignty it is your data, your rules.