Project · core infrastructure
Homelab Infrastructure
Distributed compute fabric for local AI, automation, observability, media workflows, and experimental applications.
Overview
My homelab infrastructure provides a distributed compute environment used for AI experimentation, automation systems, monitoring services, and application development. The environment is designed to function as a miniature production platform where new systems can be designed, tested, and operated across multiple nodes.
Technology Stack
Capabilities
- Distributed compute environment for AI experimentation
- Local LLM inference infrastructure
- Containerized service hosting
- Foundation for automation and agent workflows
- Platform for rapid systems experimentation
Implementation
The infrastructure combines compact compute systems and edge devices connected through a private network. Nodes run containerized workloads and participate in shared automation workflows with reproducible deployment patterns and lightweight orchestration.
Problem
Running everything on a single machine limits scalability and reliability. I needed a distributed environment capable of supporting local AI inference, containerized services, automation runtimes, observability, and experimentation with distributed workflows.
Architecture
- Control Plane · Always on
- Mac Mini M4
- 24 GB unified memory
- agent orchestration + routing
- Primary AI Compute · Always on
- Dell Pro Max GB10
- 128 GB unified memory
- ~27–70B local LLMs
- Application Host · Always on
- Raspberry Pi 5
- web apps
- dashboards
- supporting services
- Burst / Experimental Compute · On demand
- Gaming Workstation
- Ryzen 7 7800X3D
- Radeon RX 9070 XT
- 32 GB DDR5
- benchmarking + runtime testing
- Edge Layer
- Pi Zero 2 W (InkyPi)
- Pi Zero 2 W (Pi-hole)
- Raspberry Pi 4B (DogCam)
Internet │ ▼ Router │ ▼ TP-Link Switch ├───────────────────────────┬───────────────────────────┬┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┐ │ │ │ ┆ ▼ ▼ ▼ ▼ Mac Mini M4 Dell Pro Max GB10 Raspberry Pi 5 Gaming Workstation CONTROL PLANE PRIMARY AI COMPUTE APPLICATION HOST BURST / EXPERIMENTAL 24 GB unified memory 128 GB unified memory web apps · dashboards Ryzen 7 7800X3D agent orchestration ~27–70B local LLMs supporting services Radeon RX 9070 XT · 32 GB DDR5 ALWAYS ON ALWAYS ON ALWAYS ON ON DEMAND (optional) ├───────────────────────────────┬───────────────────────────────┬───────────────────────────────┐ │ │ │ ▼ ▼ ▼ Pi Zero 2 W Pi Zero 2 W Raspberry Pi 4B InkyPi Display Pi-hole DogCam
Public-safe topology view with generalized service roles and sanitized network presentation. Dashed connector marks optional on-demand capacity.
Outcome
The homelab provides a stable platform for experimentation and system development, enabling rapid prototyping and operation of AI systems, distributed services, and automation workflows without full cloud dependence.
What's Next
- Stronger observability and metrics collection
- Distributed model serving
- Expanded service orchestration capabilities
- Public-safe topology and service visualizations