The Homelab
More compute than some startups. A 3-node Proxmox cluster with 10GbE networking, 9 TiB of ZFS storage, and enough services to keep me debugging at 2 AM on a Saturday.
Server Rack
Self-Hosted Services
Everything runs on my hardware. Because why trust the cloud when you ARE the cloud?
Proxmox VE
3-node cluster with HA and live migration
TrueNAS
9 TiB ZFS storage with snapshots and replication
pfSense
Routing, firewall, VLANs, and traffic shaping
Traefik
HTTP/3 reverse proxy with auto Let's Encrypt
Matrix / Synapse
5-worker federation with coturn TURN server
VaultWarden
Self-hosted Bitwarden-compatible password vault
Headscale + Tailscale
Zero-trust mesh VPN with self-hosted control plane
Home Assistant
Smart home automation — Zigbee, Matter, Z-Wave
Jellyfin
Media server with hardware transcoding
Bazarr
Automated subtitle management
GitLab CE
Private Git + CI for personal projects
Ollama
Self-hosted LLM inference (Llama, Mistral, etc.)
AI-Augmented Workflow
I treat AI tools like I treat my infrastructure — configure deeply, integrate everywhere, and always have a self-hosted fallback.
Daily Drivers
Claude Code
Primary AI pair programmer for complex architecture work
GitHub Copilot Agent
Inline suggestions and multi-file agent mode
Cursor
AI-native IDE with custom model routing
Ollama (Self-Hosted)
Local inference for privacy-sensitive tasks and experimentation
Frontier APIs
Claude Opus/Sonnet, GPT-4o for production-grade output
MCP Integrations
Grafana MCP
Query Prometheus, Loki, Tempo, ClickHouse from AI agents
Jira/Atlassian MCP
Create issues, search tickets, update sprints via AI
Odoo MCP
ERP operations — invoicing, inventory, CRM from agent context
Bitwarden MCP
Secure credential lookup during automation workflows
Security-First AI
OWASP LLM Top 10 compliance. Prompt injection guards, PII output filtering, and hybrid inference — sensitive tasks stay on self-hosted Ollama.
# Why run a homelab? $ echo "Because every production outage I've fixed started with something I first broke at home." # The homelab is where I: $ cat reasons.txt - Test Kubernetes upgrades before touching production EKS - Prototype observability patterns (Prometheus, Loki, Tempo) - Experiment with AI/ML inference on real hardware - Run my own comms (Matrix), security (VaultWarden), and DNS - Prove that self-hosting saves money AND teaches more $ uptime 3:47am up 847 days, 14:23, load average: 0.42, 0.38, 0.31