Revisiting MetalGraph: What I Learned About MLX, Replay, and Sidecar PerformanceMay 11, 2026·16 min read
dgx-spark-inference - Keeping good habits around local inferenceLocal inference has a habit of becoming folklore. You launch a model from a shell history fragment you can never quite find again. A context setting lives in a note somewhere. A faster quantization geJul 1, 2026·8 min read
MobileLLM-R1-950M meets Apple SiliconFrom Stub to Coherent MLX Model, a story-shaped release note Getting Meta's efficient 950M parameter model running natively on Apple Silicon for fast, local inference I’ve done a tiny bit of contributions to the llama.cpp community, including adding ...Sep 15, 2025·6 min read
vault-mcp: A Scrappy, Self-Updating RAG Server for Your Markdown HoardTL;DR: I built a tiny server that watches my Obsidian vault, re-indexes only what changed, answers questions—and exposes itself as both a REST API and an MCP server so Claude can call it directly. Grab the repo: https://github.com/robbiemu/vault-mcp ...Aug 30, 2025·8 min read
Four Models, One Prompt: Who Writes the Best Instructions for AI?I Had Four LLMs Compete to Write the Perfect Prompt. The Results Were a Masterclass. I hesitate to use wording like “masterclass” especially when it is Gemini 2.5 Pro, and not some other person, telling me to. And also because I don’t usually use wor...Aug 2, 2025·13 min read
Autoblogger: Automating Tutorial Blogs with AI Agents and vibe codingI built Autoblogger after realizing how tedious it would be to write up a technical tutorial by hand on a project of mine. My recent side project, zsh-tips-agent (an AI-driven shell-tips tool), was a fun experience to write because I basically just d...May 30, 2025·7 min read
How to Build zsh-tips-agent: A Step-by-Step JourneyThis tutorial was generated with autoblogger by analyzing the git history of the project. GitHub url: https://github.com/robbiemu/zsh-tips-agent Project Overview This project is a Zsh enhancement tool called zsh-tips-agent that provides personalized ...May 30, 2025·44 min read
On KL-Divergence and Context Size Optimization in GGUF Quantizationnote: This paper was written with AI, and the exploration it describes was done collaboratively with AI. The "we" described here is us; me and a few models. With llama.cpp model quantization, properly adjusting models to keep their performance after ...Oct 28, 2024·13 min read
