HomeLab — Self-Hosted Infrastructure

Infrastructure playground used to host AI experiments and production side projects. Runs 15+ services 24/7 on a recycled notebook with enterprise-grade reliability.

Role: Creator
Year: 2026
Stack: Proxmox · Docker · Dokploy · Cloudflare · Ansible · Terraform · Linux

Context

Running production-grade services at home requires robust infrastructure. Most homelabs are either overengineered or unreliable. Needed a setup that could host client demos, personal projects, and CI/CD pipelines with enterprise reliability, while avoiding cloud costs and vendor lock-in.

Approach

Designed a Proxmox-based virtualization cluster on repurposed hardware with LXC containers.

Notebook-based: HP Pavilion dv6 (8GB RAM) running 15+ services 24/7

Hypervisor: Proxmox VE (VM on notebook)

Container Orchestration: Portainer (stable services) + Dokploy (new deployments, host mode)

Networking: Cloudflare Tunnel (zero open ports, TLS termination) + Tailscale VPN (admin access)

Observability: Prometheus + Grafana + Jaeger + OpenTelemetry Collector + Loki + Tempo

Infrastructure as Code: Terraform for Cloudflare DNS, Ansible for container management

Self-Automation: Custom MCP server enabling AI agents to manage infrastructure directly

Every architectural decision is registered as an ADR (Architecture Decision Record) to track reasoning, tradeoffs, and future migration paths.

Outcome

15+ services running 24/7: Including Portainer, N8N, Dokploy, n8n workflows, Media stack (Jellyfin, qBittorrent), Observability stack (Grafana, Prometheus, Jaeger, OTEL Collector, Loki, Tempo), Documentation (MkDocs)

99.9% uptime: Single notebook setup with comprehensive backup and monitoring

10k+ traces/month: Distributed tracing for AI agent sessions via Jaeger

Real-time alerting: Prometheus + Alertmanager routing to Telegram bot via n8n webhook automation

Zero attack surface: Cloudflare Tunnel routes only necessary services to the internet

Security hardening: Custom MCP server scoped to containers (not root), allowlist-based operations, audit logging

Documentation as truth: Every decision documented in ADRs, security audits published

Learning platform: Self-hosted infrastructure for AI experiments, automation workflows, and production projects

🔗 Production notebook →🔗 AI observability guide →