Three small computers, one real experiment
A governed, bilingual remote computing system for cosmic-ray physics in Mérida, Venezuela, built on hardware most people would throw away.
What a lab in Mérida has
LAGO, the Latin American Giant Observatory, studies cosmic rays with water-Cherenkov detectors across the continent. Its site in Mérida, at the Universidad de Los Andes where I studied, needed computing that researchers and students could reach from anywhere. The hardware available was three Mac mini units from 2014.
I turned them into a governed system: Ubuntu Server, persistent runners registered to a GitHub organisation so that jobs are submitted as code and reviewed in the open, workspaces wiped between jobs because the soldered memory cannot be upgraded, and a hardening configuration kept as a documented drop-in. Everything is written twice, in Spanish and English.
Exhibit 1DiagramRead left to right
Three small computers, one real experiment
Old machines become useful when the system around them is governed: jobs arrive as code, run in a clean workspace, and leave results and logs that anyone in the group can read.
Why a small system matters
Along the way I fixed a blocking bug in the Geant4 simulation of the detector, where photons from muon decays were contaminating the signal, and shipped a repair release that restored safe kernel defaults. Each fix is a tagged release with a short explanation, so the next person can follow the reasoning.
Venezuela’s presence in the Latin American strategy for high-energy, cosmology and astroparticle physics is, in practice, LAGO. A documented, working system, however small, is an argument in that process: it shows what a group can run and govern on its own, and what it could do with more.
Exhibit 2NumbersEach figure links to the record that states it
The system, in numbers
Exhibit 3ResourcesOpen the thing itself, not a description of it