AI & Local Models
Stop Paying for Copilot: Run Code-Llama Locally in VS Code
Ditch the subscription. A quantized Code-Llama model running locally rivals Copilot for completions—no keystrokes sent to a server. Here's the full setup w
Hands-on writing about local AI models, the software built on them, and the machines you run it all on.
Operating Systems
i3 vs Sway: Tiling Window Managers for Developer Productivity
Set up i3 and Sway tiling window managers for a keyboard-first dev workflow. Practical guide to configuration, workspaces, and choosing X11 or Wayland.
Peripherals & Gadgets
Mechanical Keyboard Switch Break-In: Does It Actually Change the Feel?
The science and ritual of wearing in mechanical switches. Smoothing, spring ping, and whether your fingers are just getting used to the scratch.
AI & Local Models
Llama 3.1 8B Quant Shootout: 4‑bit to 6‑bit on RTX 3060
On an RTX 3060, Q5_K_M runs Llama 3.1 8B nearly as fast as 4-bit with better output. See the numbers for Q4, Q5, Q6.
AI & Local Models
Local PDF Q&A Without Hallucinations: Ollama + LangChain
Build a local document Q&A system that doesn't lie. Exact chunking strategies, retrieval setup, and prompt constraints to stop LLM fabrications cold.
AI & Local Models
Local Speech-to-Text with Whisper: Faster-Whisper vs whisper.cpp for Dictation and Subtitles
Compare Faster-Whisper vs whisper.cpp for local speech-to-text. Dictation vs subtitles: setup, speed, hardware trade-offs — all offline, private.
Self-Hosting
Stop Running AI on Your Laptop: Build a $24 Cloud Inference Server
A sleeping laptop kills your model. Run Ollama on a cheap DigitalOcean droplet for always-on inference—$200 free credit with referral link.
AI & Local Models
Running Open-Source LLMs on a $300 Mini PC: Intel N100 vs. Ryzen 5 with Ollama
Intel N100 vs Ryzen 5 5500U for local LLMs on a $300 budget: token speeds, memory limits, and which mini PC keeps the conversation flowing with Ollama.
AI & Local Models
RTX 4070 GPU Showdown: Llama 3.1 8B vs Mistral 7B vs Phi-3 Medium
Head-to-head local LLM benchmarks: Llama 3.1 8B, Mistral 7B, and Phi-3 Medium on RTX 4070 12GB. Tokens/sec, VRAM, and quantization trade-offs—see which mod
Hardware
The SMR Drive Trap: Slow ZFS and How to Avoid It
Shingled magnetic recording drives can silently cripple ZFS pools with timeouts, faults, and multi-day resilvers. Here's how to spot and avoid them.
Self-Hosting
Self-Host Your Bookmarks with Linkwarden: Beat Link Rot
Set up Linkwarden to self-host bookmarks with automatic page archiving. A step-by-step guide to a permanent, private archive you control on your own hardwa
Homelab
NUT UPS Shutdowns: USB Passthrough & Graceful Script
Get your Proxmox host to shut down cleanly before the UPS battery dies. NUT config, USB passthrough, and a shutdown script that actually fires.
Self-Hosting
Build a Private Search Engine with SearXNG
Ditch tracking and ads by self-hosting SearXNG, a metasearch engine that aggregates results privately. Step-by-step guide for your homelab.