Guides

Step-by-step builds for running AI, automation, and services on your own hardware. Most have a matching short up top.

homelabprivacyai

Turn an old webcam into an AI security camera with Frigate

Run Frigate NVR on a spare PC to detect people, cars, and animals in real-time from cheap IP cameras or old USB webcams. Local processing, no cloud, integrates with Home Assistant.

homelabprivacyai

Turn a $15 Android TV Box into an Armbian Linux Server

Flash Armbian on Amlogic S905/S922 TV boxes to run Docker, Pi-hole, and other services. Boot from USB first—eMMC flashing can brick it.

homelabprivacy

Turn an Old Laptop into a Home Server with CasaOS

Install CasaOS on Debian/Ubuntu with one command and get a click-install app store for Jellyfin, Nextcloud, Home Assistant. Runs on x86 and ARM.

aihomelabprivacy

Run Local AI on Old NVIDIA/AMD GPUs with Ollama

Dust off that graphics card and run Llama, Gemma, Qwen locally with Ollama. Pick a model that fits your VRAM, skip the hype.

homelabprivacy

Receive weather satellite images with a $30 RTL-SDR dongle

Build a Meteor-M2 LRPT weather satellite receiver using SatDump, an RTL-SDR dongle, and a simple V-dipole antenna. Decode live weather imagery on your homelab.

homelabprivacy

Build a Tiny Windows 11 ISO for Old Hardware

Use Tiny11 Builder to strip bloat from official Windows 11 images and create a lean installer that runs on older PCs. You keep your license; it's just removal, not cracking.

homelabprivacy

Run Big Sur and Newer on Unsupported Macs with OpenCore Legacy Patcher

Use OpenCore Legacy Patcher to install macOS Big Sur through Sequoia on Macs Apple says can't update. Supports Penryn and newer; some GPU features may break.

homelabprivacy

Live plane tracking with a $30 RTL-SDR stick and tar1090

Decode ADS-B broadcasts from aircraft overhead using a USB DVB-T tuner, readsb or dump1090-fa, and tar1090's web interface. See callsigns, altitudes, and speeds on a live map.

homelabprivacy

Turn an Old Wi-Fi Router into a Travel VPN with OpenWrt

Flash OpenWrt on a supported router, configure WireGuard as a VPN client, and tunnel hotel or café Wi-Fi back to your home network securely.

homelabprivacy

Run a Homelab Server on an Old Android Phone with Termux

Turn spare Android hardware into a low-power compute node. Install Termux, add packages, and run services behind your LAN—no cloud rent, no monthly bills.

homelabprivacy

Build a DIY 16-bit Magnetic Core Memory Module

Hand-wind and assemble a functional magnetic core memory array using the SAM Computer reference design. Learn retro computing hardware from the ground up with ferrite cores and wire.

homelab

Turn a $30 TV Tuner Stick Into a Live Plane Tracker (ADS-B)

Map the aircraft over your house with an RTL2832U USB stick or RTL-SDR Blog V4, a Raspberry Pi or old laptop, readsb or dump1090-fa, and tar1090. Listen-only ADS-B on 1090 MHz.

aihomelabprivacy

Run AI Models on PS5 with llama.cpp GGUF

Turn your PlayStation 5 into a local AI inference engine using llama.cpp's native GGUF support. Run Llama, Qwen, and other quantized models directly on console hardware for private, offline inference.

homelabaiprivacy

Turn Old WiFi Gear Into a 320Msps SDR Receiver with GQRX

Repurpose abandoned WiFi hardware as a software-defined radio receiver. Build a 320Msps SDR station on your homelab using GQRX, GNU Radio, and RTL-SDR dongles—no new gear required.

aihomelabprivacy

Run MiniMax H3 in ComfyUI: Local Audio & 2K Video

Generate audio and 2K video together on your own hardware using MiniMax H3 in ComfyUI. No API costs, full control, offline workflows.

aihomelabprivacy

Self-host Milvus: ditch expensive vector DB SaaS

Run Milvus standalone on your homelab to store and search vectors at scale. Skip the cloud subscription and own your AI infrastructure completely.

homelabprivacyai

Run macOS Binaries on Linux ARM with Kakehashi

Use Kakehashi to execute native macOS applications on Linux ARM64 hardware, cutting Apple hardware costs while keeping your workflow intact.

aihomelabprivacy

Run OpenHands Agent Canvas on Your Own Hardware

Deploy OpenHands Agent Canvas on bare metal to run AI coding agents locally, keeping compute and data on your own server instead of renting cloud resources.

homelabprivacyai

Build Hister: Your Own Private Search Engine

Index your local files and browser history in a self-hosted search engine. Keep your search data private, searchable, and under your control—no cloud, no tracking.

homelabprivacy

Own Your Hypervisor: KVM on Linux Instead of VMware

Ditch VMware licensing hell. Build a KVM hypervisor on commodity hardware, get full control over virtualization, and run your homelab on your own terms.

homelabprivacyai

Rotary Phone over LTE: ESP32-S3 + Asterisk PBX

Wire a vintage rotary dial to an ESP32-S3 to make real phone calls over LTE via Asterisk. Retro hardware, modern VoIP backbone, no carrier needed.

homelabprivacyai

Run Organic Maps Offline for Disaster Response

Deploy Organic Maps on local devices for navigation without cellular or internet. Works entirely offline with OpenStreetMap data, critical for emergency response and remote operations.

homelabprivacy

Install postmarketOS on your phone for full Linux control

Replace Android with postmarketOS to run a real Linux distro on your phone. Full control, privacy, no vendor lock-in—but plan for hardware limits and community support.

aihomelabprivacy

Run 104GB Qwen on 48GB Mac with Ollama

Fit massive quantized models on consumer Mac hardware using Ollama's efficient inference. Get 12 tokens/sec with strategic quantization and memory management.

aihomelabprivacy

Run 70B LLMs on 4GB GPUs with AirLLM

Use AirLLM's smart memory management to run enterprise-scale language models on consumer GPUs. Fit 70B parameters on 4GB VRAM by streaming weights on-demand instead of loading everything upfront.

aihomelabprivacy

Quantize Bonsai 2 27B to 4-bit with AutoGPTQ

Use GPTQ weight-only quantization to run a 27B LLM on 9x less VRAM, keeping quality intact. Compress locally, inference on modest GPUs.

aihomelabprivacy

Run Claude Code and Codex Locally with OtoDock

Deploy OtoDock on your server to get local code generation without cloud API costs. Self-hosted alternative to Claude Code and GitHub Copilot.

homelabprivacyai

Control a Dumb AC Unit with ESP32 and ESPHome

Wire an ESP32 to your window AC's remote control pins, then command it over your LAN using ESPHome. No Wi-Fi remote needed, total cost under $35.

homelabprivacy

Build a Local Game Library with EmulationStation

Skip Game Pass rentals. Set up EmulationStation on bare metal to own and play classic games from your own hardware, no subscriptions required.

aihomelabprivacy

Run Claude Code Locally with OtoDock

Deploy OtoDock on your own hardware to run Claude Code and Codex as local AI agents without paying per API call. Full self-hosted setup.

aihomelabprivacy

Self-host Claude Code agents with OtoDock

Run Claude's code execution engine on your own hardware. Replace the SaaS with OtoDock—a self-hosted agent framework that keeps your inference and execution local.

homelabprivacyai

Self-host Kubernetes: escape the cloud tax

Run Kubernetes on your own hardware to replace managed cloud services. Deploy and scale containerized apps without paying AWS, Azure, or GCP monthly bills.

homelabprivacy

Archive YouTube channels with yt-dlp on bare metal

Set up automated YouTube video downloads on a homelab server using yt-dlp. Build a self-hosted archive that respects creators and keeps your content offline.

aihomelabprivacy

Run LLMs on Minimal Hardware with llama.cpp

Use llama.cpp to run quantized language models on constrained hardware. Optimize inference with GGUF formats and CPU backends for your homelab.

aihomelabprivacy

Tune Ollama for Real Performance: Memory, Quantization, and Hardware Hacks

Stop running LLMs at half-speed. This guide covers quantization levels, context windows, GPU offloading, and CPU tuning to squeeze real inference gains from Ollama on your own hardware.

homelabprivacyai

Run rust-analyzer on your homelab for 100x less RAM

Deploy a self-hosted Rust LSP server on cheap hardware. Get IDE features—completion, refactoring, diagnostics—without melting your machine. Keep it on your LAN.

aihomelabprivacy

Run Ollama Locally: Skip Google's Cloud, Own Your AI

Deploy Ollama on bare metal to run open LLMs like Gemma 4 without API costs or cloud lock-in. Chat, code, and build with models on your own hardware.

aihomelabprivacy

Build Your First AI Agent with LangChain

Set up LangChain on bare metal to chain LLMs with tools and data sources. Run agents locally without cloud dependencies.

aihomelabprivacy

Run Qwen 80B on Mac with 4.3GB RAM using llama.cpp

Quantize and run large language models locally on Apple Silicon with minimal memory footprint. No cloud subscriptions, no API costs, full privacy.

aihomelabprivacy

Run Open WebUI for Private AI Chat on Your Hardware

Self-hosted AI companion using Open WebUI, Ollama, and local models. Full control, no cloud, works offline on your homelab.

aihomelabprivacy

Run Private AI Chat Apps with Ollama

Self-host conversational AI on your hardware using Ollama and community chat interfaces. No cloud, no API keys, full control over your models and data.

aihomelabprivacy

Self-host AutoGPT: build and run AI agents locally

Deploy AutoGPT on your own hardware to build, manage, and execute AI agents without cloud dependency. Control infrastructure, bring your own models, run agents on-demand or scheduled.

aihomelabprivacy

Run Stable Diffusion Locally on Bare Metal

Self-host text-to-image generation with Stable Diffusion v1 on your own GPU. Generate images from prompts without cloud APIs or subscriptions.

aihomelabprivacy

Run Fruit AI locally for video generation

Self-host Fruit AI on bare metal to generate fruit-themed videos without cloud dependencies. Keep it on your LAN.

aihomelabprivacy

Run Local LLMs on Bare Metal with Ollama

Install Ollama on Linux hardware, pull open models like Gemma, and serve them via REST API for private AI without cloud costs or data leakage.

aihomelabprivacy

Build AI Prompts for Business with Ollama

Run Ollama locally to generate business prompts—sales copy, emails, proposals—without cloud costs or API fees. Self-hosted, on your hardware, behind your network.

aihomelabprivacy

Run GLM 5.2 on AMD MI355X with vLLM

Deploy Alibaba's GLM 5.2 model on AMD MI355X GPUs using vLLM for 2600+ tokens/sec local inference with PagedAttention optimization.

aihomelabprivacy

Run Bonsai 27B Locally with Ollama

Get a capable 27B parameter AI model running on your hardware with Ollama. Pull, run, and integrate Bonsai into your homelab for private inference.

homelabprivacy

DIY Bowling Alley Scoring with ESP32 and ESP-IDF

Replace enterprise bowling systems with ESP32 microcontrollers running ESP-IDF. Build a networked scoring system for your homelab at a fraction of commercial costs.

aihomelabprivacy

Run LLMs Locally with Ollama: Complete Self-Hosting Guide

Set up Ollama to run state-of-the-art language models on your own hardware. No cloud fees, full privacy, complete control.

privacyhomelab

Self-Host a Password Manager: Vaultwarden in 10 Minutes

Self-host Vaultwarden, a lightweight Bitwarden-compatible password manager, in 10 minutes. Docker setup, HTTPS done the safe way, and no public exposure.

privacyhomelab

Self-Host Immich: The Photo Backup That Kills Google Photos

Self-host Immich to replace Google Photos — automatic phone backup, AI-powered search, and off-site copies to Backblaze B2, all on hardware you own.

aihomelabprivacy

Run a Local LLM on a Mini PC: Self-Host Your Own ChatGPT

Run a private ChatGPT-style AI on a cheap mini PC with no GPU. Full Ollama + Open WebUI setup, real speed numbers, and which models actually fit.

n8nhomelabprivacy

Self-Host n8n and Replace Zapier: Setup + Real Cost Breakdown

Replace Zapier with self-hosted n8n in 10 minutes. Docker Compose setup, first workflow, and the real monthly cost math versus Zapier and Make.