· field notes
Notes on firmware, signals & ship-it.
Long-form write-ups from the bench — local AI, embedded systems, debugging war stories, and whatever I learn the hard way.
Gemma 4 12B: multimodal AI that runs on your laptop
Google's new encoder-free open model runs text, image, audio and video locally on 16 GB — here's what's new and how to try it.
Getting started with local AI
What local LLMs are, why run them yourself, the hardware basics (RAM/VRAM, quantization), and how to pick a model.
How to use Ollama
Install Ollama, download and chat with a model in one command, and call it from your own apps via the local API.
Using models in LM Studio
Install LM Studio, browse and download GGUF models in the GUI, chat, and run a local OpenAI-compatible server.