Can integrated GPUs accelerate local LLMs? I explored running self-hosted models like Ollama with ROCm on an AMD 5600G-based homelab, but ran into kernel and memory allocation limitations. Here’s what I found and where things stand today.

What to do when pip can’t install a library from a Git source due to flat layout issues—and how Poetry, setuptools, and PEP 621 come into play.

Solving 4 years of technical debt and adding containerisation to the project.

My workflow for creating notes in Obsidian on a Mac, starting with the contents of an email in Mail.app.

Publishing a Jekyll Site to GitHub Pages using GitHub actions has got a lot easier than last time I looked into this.