Antonio Ruvalcaba · Vol. III · № 47
a notebook of machines & minds
— MANIFESTO FOL. I · RECTO
Computer Scientist
AI Infrastructure

on machines
& the minds
that coax them

A working notebook. What I build, what I read, the quiet hours that make sense of them. From first principles.
OBSERVATORY / COMPOSE FIELD: NOISE
0000·0000
+ N 40°25' W 003°42' + + ƒ = logo τ = 0.00 +

There is a quiet work that happens before anything is said out loud.

Before the video, the essay, the post, before the finished thing exists at all, someone sits in a private room with a problem and takes it apart. Traces it to the root. Writes it down for themselves, slowly, until it stops looking clever and starts looking true. Most of what I want to do in the world happens in that room.

This site is the window into it.

I'm a computer scientist writing from Mexico City. I've always been fascinated by two things at once: human behavior, and the tools humans build. Of all the tools, the computer is my favorite. What keeps me coming back to it is the reciprocity: how we build with computers, and how they build us back.

The line from the assembler to the compiler to Unix is a line of tools, each one changing what it meant to program a computer. AI is the newest chapter in that lineage, and the one I want to sit with longest. I care about all of it: the building, the watching, the pros and cons, the way each piece of it actually gets shipped into the world.

The same instinct pulls me toward where the disciplines cross. A theorem from the eighteen-hundreds turns up inside a modern compiler. A question Leibniz was asking in the seventeenth century, whether reasoning itself could be reduced to calculation, is open again inside modern AI research. Mathematics and philosophy, in the end, are both about looking patiently at what is actually there.

My attention right now is on artificial intelligence, and on the older question it keeps pushing back into view: what a mind is, and what sort of life is worth building around one. What interests me is the craft: the careful, patient, hands-on work of understanding a system well enough to teach it.

What I'm trying to do, ultimately, is simple. Learn computer science, mathematics, and philosophy well. Teach what I learn so carefully that the explanation still works in ten years. As AI makes most technical writing disposable, that "still works in ten years" is the only bar worth caring about.

The rest is just the daily work. Study the foundations. Build things that work. Write down what I learn. Assemble a life that holds together. Enjoy, for its own sake, the work of figuring out how the world works.

Tending
An adaptive retrieval layer for a personal brain
TF-IDF, embeddings, and co-activation edges that strengthen with use. A notebook that reorganizes itself around the questions I ask it.
Building
Operand, a polyglot MCP orchestrator
One Go binary that spawns tools in any language as child processes. Writing the runtime I wanted for my own agents.
Writing
This notebook, in public
Org-mode files, a hundred lines of Elisp, a GitHub Action. The site built to enforce the writing habit.
No dispatches yet. The Ledger is quiet.
“The unexamined program is not worth running.” — after Socrates, with apologies
Set in
EB Garamond & JetBrains Mono
Hand-drawn
in Flask, served from fly.io
Last tended
April 2026
License
words CC-BY · code MIT