Hi, I'm Antonio Ferreiro

P-01

cryptoKit

A Java desktop application to encrypt, decrypt, hash and sign with 44 algorithms, from Caesar and Enigma to post-quantum ML-KEM, and a site that teaches the history and mathematics of cryptography.

• 5 stars
• Updated Oct 2026
• Since 2017
• GPL-3.0

My oldest project (2017), rewritten as version 2.0 (2026). The Java Swing application, with light and dark themes and in English and Spanish, brings together classical ciphers, encodings, hashes, HMAC, password hashing (Argon2id, scrypt, bcrypt), authenticated symmetric encryption (AES-GCM, ChaCha20-Poly1305), public-key cryptography (RSA, X25519) and the post-quantum algorithms ML-KEM and ML-DSA; the same jar is also a command-line tool. It comes with a portal of ten chapters on history and mathematics, from Caesar to lattices, with theorems and interactive figures that run the real algorithms in the browser.

Key decisions

• Every algorithm carries a security badge (recommended, legacy, broken…) and a “How it works” link to its chapter on the portal.
• Safe defaults: random IVs and nonces, authenticated modes that reject tampered data, and constant-time comparisons.
• 89 tests with official vectors (NIST, RFC) and with ciphertexts made by version 1.0, which still decrypt.
• Released with GitHub Actions: the jar and portable packages for Windows, macOS and Linux with their own Java; the portal on GitHub Pages.

• Java 21
• Swing
• FlatLaf
• Bouncy Castle
• Maven
• JavaScript
• Python
• GitHub Actions

• Java 100%

P-03

Wahoo

A complete compiler, written in Rust, for a language from the Mushroom Kingdom that targets WebAssembly, and a site that teaches how a compiler works.

• 0 stars
• Updated Oct 2026
• Since 2026

Wahoo is a statically typed language (coin, pipe, bounce, wahoo(…)) and its compiler, written in Rust with no dependencies: lexing, parsing, semantic analysis, optimisation and WebAssembly generation with a hand-written binary encoder. The site explains each phase in seven chapters with interactive figures and a playground that runs the real compiler in the browser, itself compiled to WebAssembly.

Key decisions

• Every classic phase written by hand, with no parser generators or Wasm libraries: recursive descent, Pratt parsing and error recovery.
• Errors in English and Spanish with exact positions, hints and "did you mean…?".
• The same compiler runs in the terminal (on the wasmi interpreter) and in the browser, compiled to wasm32 with a tiny C ABI.
• In CI, every example program is compiled, run and compared with its expected output, with and without the optimiser.

• Rust
• WebAssembly
• wasmi
• Python
• JavaScript
• GitHub Actions

• HTML 37%
• Rust 28%
• JavaScript 21%
• CSS 9%
• Python 6%

P-05

Teaching sites

Four open sites to study software architecture and the mathematics behind AI, quantum computing and computer science.

• 12 stars
• Updated Oct 2026
• Since 2026

What I study, I publish. Architect Study Hub is my architect's notebook; Math of AI, Math of Quantum and Calculus × Computing explain the mathematics of modern computing with theorems, history and interactive figures. All of them are static sites published with GitHub Actions on GitHub Pages.

• Architect Study Hub

  A knowledge base on architecture, platform, AI and cloud, with 200 interview questions, a timed mock interview and an AI study assistant that uses the page you are reading as context.

  MkDocs · Python · LLMs4Code

• Math of AI

  Why AI works: from circuits and the Turing machine to LLMs, with theorems, history and demos (including a Prolog interpreter in the browser).

  Dependency-free Python · MathML4Code

• Math of Quantum

  The mathematics of quantum computing, from qubits to Shor and Grover, with proofs and a state-vector simulator.

  Dependency-free Python · MathML4Code

• Calculus × Computing

  A navigable map of university calculus and where each idea shows up in AI, graphics, simulation and robotics.

  Astro · TypeScript0Code

Key decisions

• Theory first: theorems with foldable proofs, and an interactive figure wherever a formula alone is not enough.
• Static sites with no backend, no analytics and a strict CSP; the maths is converted to MathML at build time.
• Three different stacks on purpose: MkDocs, a dependency-free Python generator, and Astro with TypeScript.
• Continuous deployment with GitHub Actions; the maths sites are in English and Spanish.

• Python
• MkDocs
• Astro
• TypeScript
• MathML
• GitHub Actions