Chess AI

A fully playable chess game with an AI opponent, built from scratch in Python using Pygame.

Features

• Complete chess rule implementation: legal move generation, check/checkmate/stalemate detection, castling, en passant, pawn promotion
• AI opponent using Principal Variation Search (PVS) — an optimized variant of minimax with alpha-beta pruning
• Custom positional evaluation function using piece-square tables for every piece type
• Move ordering heuristics to improve search efficiency:
  • Checkmate-move prioritization
  • Capture-move prioritization
  • History heuristic
  • Killer move heuristic
• Pawn structure evaluation — isolated pawns, passed pawns, and pawn chains are factored into the AI's positional judgment, not just material count
• Move animation, move log display, and sound effects
• Runs AI search in a separate process (multiprocessing) to keep the UI responsive while the engine "thinks"

Why this project

Built to deepen my understanding of search algorithms, game state management, and performance optimization — independent of my data analytics work. It's a from-scratch implementation, not built on a chess library, including hand-written legal move generation and a custom evaluation function.

How to run

pip install -r requirements.txt
python chess_main.py

Controls

• Click a piece, then click a destination square to move
• Z — undo last move
• R — reset the board

Tech

Python, Pygame, multiprocessing (for non-blocking AI search)