Javi Falcón
ESEN

I love computers. Right now I have the time and, thanks to AI, the tools to take on projects that would once have been beyond me.

This is what comes out of it. Most of my hours go into Falcon, a C++ chess engine with neural-network (NNUE) evaluation that competes on Lichess. Around it there are games you can play right in your browser, a web app demo built with Laravel, and the code I share on GitHub.

Knight's tour: all 64 squares, none twice.
  1. Projects

    Engines, games, tools and demos, each with its own page.

  2. Blog

    Notes on chess, AI and programming.

  3. Play

    Games you can play without leaving the browser.

  4. Now

    What I'm up to lately.

Featured

See all projects →

From the notebook

  • Losing to a linear network

    In the second half of September I lost every game that mattered. Not me: my networks. Every neural network with hidden layers I trained for Falcon, my chess en…

    · 8 min

    • falcon
    • nnue
    • chess
    • methodology

RSS →

Play

  • Pipe-Muse

    A playable tribute to Pipe Mania (1989): lay the pipeline before the flooz arrives and cover the minimum distance without spilling. 10×7 gr…

    b7 · Game · TypeScript

    Code →Play →

  • Nibbo

    A browser maze arcade game: Nibbo, a lantern critter, roams a hedge garden eating light seeds while four shadow moths hunt it, each with it…

    c3 · Game · JavaScript

    Code →Play →

  • Sudoku

    Classic 9×9 Sudoku in a single index.html file: no frameworks, no CDN, 100% offline. Every puzzle is genuinely generated from the seed, wit…

    g6 · Game · JavaScript

    Code →Play →

Now

Updated: 7 Oct 2026

  • Training Falcon’s network on real game results, after discovering the data I was using had none.
  • Writing a series about how I trained that network; the first article is already published.
  • Redesigning this website.