local models | hardware |
bad ideas | worse projects
Full-time developer experimenting with local AI, coding agents, hardware, and token efficiency. Using this space to document the models I run, the tools I build, and the configurations I try along the way.
Blog Posts
First Steps: The tokenBuffalo Rig (ft. AliExpress)
My work is full-time coding. My colleagues are increasingly AI agents, so I did the reasonable thing and built them a server.
This is the first version of the tokenBuffalo rig: a used Xeon, 96 GB of ECC memory, two RTX 3060s, and a suspiciously cheap X99 motherboard. This post covers the build, the hardware choices, the setup, and what it can actually run.
tokenBuffalo 