Working-draft book teaching category theory as an engineering tool through a typed Rust machine learning pipeline, with runnable examples and a public GitHub repo.
Key Takeaways
Domain objects map to Rust types, morphisms to typed transformations, and composition to executable program structure, making category theory concrete rather than decorative.
Authors are Hamze Ghalebi (Paris-based AI architect, Remo Lab) and Farzad Jafarranmani (PhD, Université Paris Cité, Lagrange Mathematics and Computing Research Center / Huawei).
Book is intentionally published as an incomplete draft; chapters, code, and diagrams are still evolving and public feedback via GitHub issues is explicitly requested.
Free to read online permanently at hghalebi.github.io/category_theory_transformer_rs; commercial or organizational reuse beyond individual study requires written permission.
A public workshop hosted through AI Reading Club accompanies the draft as a live study format.
Hacker News Comment Review
One commenter flagged that “ML” should be expanded to “Machine Learning” to avoid confusion with the ML programming language, a real naming collision in a category-theory context where the ML language is historically relevant.
Notable Comments
@ctenb: Recommends spelling out “Machine Learning” to avoid ambiguity with the ML programming language.
