Build your own competitive chess engine over 6–8 weeks! This project takes you from game theory fundamentals through classical AI search all the way to neural network evaluation (NNUE) — the same technology powering engines like Stockfish.
- Game theory fundamentals — zero-sum games, minimax, adversarial search
- DFS / BFS and building solvers (Tic-Tac-Toe & Notakto)
- Minimax algorithm with Alpha-Beta Pruning & move ordering
- Chess library integration — board representation & move generation
- Zobrist hashing, transposition tables & quiescence search
- Iterative deepening & time management
- NNUE (Efficiently Updatable Neural Networks) — reading, understanding & implementing
Expect ~5 hours/week. It's a bit hectic but super rewarding!
- Language: C++ (engine core), Python (optional RL/training)
- Library: Disservin Chess Library
- Protocol: UCI (Universal Chess Interface)
- Testing: CuteChess / Arena GUI
A fully playable chess engine you built from scratch — submitted to a round-robin Engine-vs-Engine tournament. Best engine wins. ♟️