Posts
Page 12 of 41
-
LangGraph vs CrewAI vs AutoGen: AI Agent Frameworks for Mere Mortals
Everyone's talking about AI agents like they'll solve world hunger by Tuesday. But which framework do you actually use? We compare LangGraph, CrewAI, and AutoGen — with working Python examples, brutal honesty, and a healthy dose of skepticism about your robot assistant booking flights to Reykjavik.
9 min read -
Systemd Timers vs Cron: Scheduling That Doesn't Suck
Cron has been scheduling your jobs since before you were born. Systemd timers do everything cron does, plus logging, dependencies, and missed-run recovery.
6 min read -
Whisper & Faster-Whisper: Self-Hosted Speech-to-Text That Actually Works
OpenAI Whisper is genuinely impressive speech-to-text — and you can run it entirely on your own hardware. Add Faster-Whisper into the mix and suddenly you've got transcription that's 4x quicker, uses less VRAM, and doesn't phone home to anyone. Here's how to set it all up without losing your mind.
9 min read -
Gitea vs Forgejo vs GitLab CE: Self-Hosted Git
GitLab CE does everything and wants all your RAM. Gitea and Forgejo run on a Raspberry Pi. Here's which self-hosted git platform actually fits your setup.
7 min read -
LiteLLM & vLLM: One API to Rule All Your Models
Your app calls OpenAI, your side project calls Anthropic, your homelab whispers to Ollama — and your codebase looks like a crime scene. LiteLLM and vLLM are the dynamic duo that puts a single sane API in front of every model you'll ever run, local or cloud.
7 min read -
Docker BuildKit: Stop Building Images the Slow Way
Cache mounts, secret mounts, parallel stages — BuildKit turns your Dockerfile from a slow linear disaster into something that actually respects your time.
6 min read -
Stable Diffusion vs ComfyUI vs Fooocus: AI Image Generation at Home
Confused by the alphabet soup of local AI image generators? We break down Automatic1111, Forge, ComfyUI, and Fooocus -- covering GPU requirements, Docker setups, model management, and which one you should actually start with based on your hardware and patience level.
14 min read -
LLM Fine-Tuning for Mortals: LoRA, QLoRA, and Your Gaming GPU
You don't need a server rack that doubles as a space heater to fine-tune an LLM. With LoRA and QLoRA, your gaming GPU can teach a language model new tricks — and we'll walk through the entire process without requiring a PhD or a second mortgage.
17 min read -
WireGuard vs OpenVPN 2026: It's Not Even Close
OpenVPN is the battle-tested workhorse. WireGuard is everything VPNs should have been from the start. In 2026, here's which one you should actually use.
6 min read -
RAG on a Budget: Building a Knowledge Base with Ollama & ChromaDB
Stop paying per-token to ask questions about your own documents. This guide walks you through building a fully local RAG pipeline with Ollama and ChromaDB — from Docker Compose to Python code — so your AI can actually know things without hallucinating them.
16 min read -
SSH CA: Finally Ditch authorized_keys
Managing authorized_keys across 10 servers is how you lose track of who has access to what. An SSH CA lets you sign keys and revoke access without touching every server.
5 min read -
Docker BuildKit: Stop Waiting for Your Images to Build
Your CI pipeline is spending 8 minutes installing npm packages. Every. Single. Build. Docker BuildKit has had the fix for years — parallel stages, cache mounts, and proper secret handling — and most people are still ignoring it. Let's fix that.
9 min read
