Infrastructure

Production Home Lab

Multi-node Kubernetes cluster, AI inference server with 4x Intel Arc GPUs, and automated CI/CD - all accessible via Cloudflare from anywhere.

12+
VMs
20+
Containers
99.9%
Uptime
~64GB
GPU RAM

Proxmox Cluster (2 Nodes + Pi Quorum)

HA virtualization cluster running VMs for Kubernetes, AI inference, and development. Raspberry Pi 5 as QDevice.

Kubernetes (3 Nodes)

Multi-node K3s cluster across Proxmox VMs. Hosts AI agent orchestration, monitoring, and internal tooling.

Intel Arc Pro B70 GPU Server

4x Arc Pro B70 GPUs running vLLM. DeepSeek V4-Flash with NVFP4 quantization at ~45 t/s.

CI/CD via Ansible + GitHub Actions

GitHub webhooks trigger Ansible playbooks for zero-downtime deploys across all services.

Cloudflare Tunnel

All external traffic routed through Cloudflare edge network with Access policies.

OpenWRT Router (Pi 4B)

Self-managed router on Raspberry Pi 4B with firewall rules and static IP allocation.

Docker + Portainer

Container orchestration for Jellyfin, Nextcloud, Uptime Kuma, VS Code Server, CasaOS.

Monitoring (Kuma + Grafana)

Self-hosted stack tracking health, uptime, and performance across all services.

Domain: dominickp.com

Custom domains routed through Cloudflare. Microservices via subdomains with Access.

The Evolution

What started as a fix for a flaky WiFi router evolved into a production-grade infrastructure. The journey: 2014 HP laptop running OpnSenseRaspberry Pi 4 with OpenWRTProxmox cluster with KubernetesIntel Arc Pro B70 GPU inference server.

Every component is version-controlled, deployments are automated through Ansible, and the entire stack is monitored 24/7. No cloud vendor lock-in - just self-owned, production-grade infrastructure.