Running LLMs Locally Without Losing Your Mind

The setup that finally made local inference practical — and the three things we got wrong the first two times.

(Full lab note to be written — this is a migrated stub so the route and listing work.)