Running a Local LLM Coding Server on MacBook Pro M5 Pro 48 GB

I recently set up a local coding AI on my MacBook Pro M5 Pro with 48 GB of unified memory — no cloud, no external API, nothing leaving the machine unless I explicitly route it there. Here is an honest account of what I tried, what crashed, and what finally worked.
The goal I wanted to run a capable coding model locally — no cloud, no API costs, no code leaving my network.