Built with Claude Code Contact Sign in →
Portfolio

AI-Driven Infrastructure
& Engineering

A personal infrastructure project designed and built entirely with Claude Code — Anthropic's AI coding agent. Spanning a Dell R730 XD hypervisor, a Mac Studio AI server, and a public cloud VPS, with 15+ self-hosted services, six production web applications, full observability, and an automated security pipeline.

5
Physical / Virtual Hosts
15+
Running Services
6
Web Applications
14+
Local AI Models
Hardware Layout
42U Server Cabinet
1U Exhaust Fan Top-of-rack ventilation
Network
UniFi Cable Modem WAN
UniFi UDM Pro SE Router · Firewall · IDS/IPS
UniFi Patch Panel 24P · Cat6
UniFi Pro HD 24 24P GbE · PoE
10U — Available
Compute
Mac Studio Shelf · 2U
Apple M-series · 64 GB RAM · AI Server
Tripp Lite SMART1500LCD 1500VA · UniFi + Mac
Test AI Rig 4U · Sliger
AMD Ryzen 9 9950X3D · NVIDIA RTX 5080
Gigabyte X870E · 64 GB DDR5
Dell R730 XD 2U
2× Xeon E5-2690v4 · 256 GB RAM
20× 1TB HDD · Proxmox VE
Eaton SMART1500RMXL2U 2U
1500VA · Rack UPS · R730 XD + Test Rig
17U — Available
Mac Studio
Apple M-series · 64 GB RAM
Rack shelf · Primary AI inference server
Ollama · 14+ models Open WebUI n8n Gitea SearXNG Kokoro TTS Uptime Kuma Qdrant RAG
Test AI Rig
AMD Ryzen 9 9950X3D · NVIDIA RTX 5080
Gigabyte X870E · 64 GB DDR5 · Sliger
Local GPU inference Model fine-tuning CUDA workloads
Dell R730 XD
2× Xeon E5-2690v4 · 256 GB RAM
20× 1TB HDD · Proxmox VE
Web Server VM NPM + Services Wazuh SIEM TrueNAS Immich Grafana · Prometheus Qdrant
Linode VPS
Public cloud · Linode
seventhirtylabs.com
Node.js auth server nginx PM2 Matomo analytics Resend email
Infrastructure
Proxmox Hypervisor
Dell R730 XD — 256 GB RAM, 24 cores. Manages all virtual machines, snapshots, and scheduled backups to TrueNAS.
Mac Studio — AI Server
Primary AI inference node running 14+ local language models via Ollama, Open WebUI, n8n workflow automation, and self-hosted Gitea.
Web Server VM
Serves all homelab web applications via nginx, with internal API routing across services through proxied /svc/ endpoints.
Security VM — Wazuh SIEM
Dedicated security host running Wazuh manager, indexer, and dashboard with 4 agents, plus automated code scanning on every Git push.
Public Cloud VPS
Linode VPS hosting seventhirtylabs.com — Node.js auth server, nginx reverse proxy, PM2 process management, and Matomo analytics.
Reverse Proxy + Storage
Nginx Proxy Manager handles SSL termination and public routing. TrueNAS provides shared storage for VM data volumes and backups.
Services & Stack
AI / Machine Learning
Local LLM Inference
Ollama running 14+ models — Gemma 4, Qwen 3.5, Llama 3, Phi-4, Devstral, CodeLlama. Open WebUI chat interface with Qdrant vector search for document Q&A and Kokoro text-to-speech.
Observability
Grafana + Prometheus
Grafana dashboards with Prometheus scraping node_exporter metrics from all four hosts. Uptime Kuma for service availability tracking and self-hosted Matomo for web analytics with per-user attribution.
Security
Wazuh SIEM + Code Scanner
Wazuh monitoring four hosts with custom alert rules (levels 3–14). Automated Gitea code scanner runs semgrep (SAST) and trufflehog (secret detection) on every push and nightly across all repos.
Automation
n8n Workflow Engine
n8n handles scheduled data workflows, Git webhook processing, and automated notifications. Deployments trigger via rsync hooks on file save — changes go live within seconds.
Networking
Multi-Layer Proxy Chain
Nginx Proxy Manager at the edge handles SSL and public routing. Internal VM nginx manages LAN access control and /svc/ same-origin proxying for dashboard API calls. UFW firewalls on all VMs.
Version Control
Self-Hosted Gitea
Gitea runs on the Mac Studio with HTTPS-only access. System-level webhooks trigger the Wazuh VM code scanner on every push. All project code versioned and tracked.
Applications Built

All built with Claude Code — single-file vanilla JavaScript, no frameworks or build tools.

OpenTerminal
Live
AI-powered web terminal supporting multiple language model providers (Anthropic, OpenAI), with shared API budget management, per-user monthly spend caps, and admin-configurable model selection.
Ops Dashboard
Live
Unified infrastructure control panel pulling live data from Prometheus, Wazuh, Proxmox, Uptime Kuma, and the Gitea code scanner — real-time health across all hosts in one view.
Macro Conditions Monitor
Live
Economic data dashboard pulling real-time indicators from FRED and BLS — inflation, employment, GDP, housing starts, and leading indicators — with AI-generated analysis summaries.
Hiring Demand Intelligence
Live
Labor market analytics with state-level employment breakdowns, job openings data, sector-by-sector demand trends, and leading indicator tracking for workforce planning.
QuantLab
Live
Financial modeling and quantitative analysis platform with live market data integration, charting, and analytical tools for equity and macro research.
Performance Marketing Report
Live
Campaign analytics dashboard for Acuity with KPI tracking, performance benchmarking across channels, trend analysis, and client-ready reporting views.
Platform & DevOps

Authentication Platform

  • Custom Node.js auth server with session-based authentication, JWT tokens, and cookie management
  • Role-based access control — admin panel with user approval workflow and per-user project assignment
  • Admin dashboard for user management, project configuration, API key storage, and login activity reporting
  • Self-hosted analytics (Matomo) with authenticated user ID tracking and campaign source attribution
  • Public registration with manual review, email notifications, and geographic login tracking

Deployment & Operations

  • Self-hosted Gitea with webhook-triggered security scanning on every code push
  • Automatic deployments — file saves trigger rsync hooks, changes go live in seconds
  • All services containerized in Docker Compose stacks across multiple hosts
  • PM2 process management for Node.js services on the public VPS
  • Multi-layer routing: public proxy → VM nginx → service backends
  • Nightly automated security scans with alerts flowing into the central SIEM
Key Technologies
Development
Claude Code JavaScript Python SQL Bash
Infrastructure
Proxmox Docker Docker Compose nginx PM2 systemd
AI / ML
Ollama Claude API OpenAI API Qdrant RAG pipelines nomic-embed-text
Data
SQLite MariaDB Prometheus OpenSearch
Security
Wazuh SIEM semgrep trufflehog UFW
Monitoring
Grafana Prometheus Matomo Uptime Kuma node_exporter
Platforms
Ubuntu Server macOS Proxmox VE Linode
Ask about this project Claude
Hi! I can answer questions about this infrastructure — what it runs, how it was built, and what the applications do. What would you like to know?