Blog
AI, local LLMs, cloud architecture, and developer tools.
I Fixed the 26B: From 2/10 to My Daily Driver at 49 tok/s (Ollama Can't Do This)
From 2 tok/s to 49 tok/s. Ollama's 3 bugs, the missing Q3 quant, 128K context for free, Qwen 3.5 35B on 24GB, and a Claude Code clone running locally. Plus: computer use with bounding boxes.
I Tested Every Gemma 4 Model Locally on My MacBook -- What Actually Works
Audio ASR in 3 languages, image understanding, full-stack app generation, coding, and agentic behavior. E2B vs E4B vs 26B vs 31B on M4 Pro 24GB. Plus free cloud API benchmarks.
The Ralph Wiggum Technique: Autonomous AI Development with Claude Code
How a goat farmer's 5-line bash script changed AI-assisted coding forever. Run Claude Code in an infinite loop while you sleep.
The Ultimate AI Code Editor Showdown
Roocline vs Aider vs Windsurf vs Cursor vs GitHub Copilot. A comprehensive comparison based on real-world usage and community feedback.
Cloud Architecture Best Practices 2026
Serverless, multi-cloud strategies, and AI-powered infrastructure management.