card-game-ai

KARDS AI simulator and reinforcement learning project for the KARDS card game: AlphaZero self-play, PUCT MCTS, Python, and PyTorch.

Multi-variant rummy simulator + evaluation framework for card-game AI (heuristic / RL / CFR / LLM / GNN / hybrid). See PAPER.md for the research programme.

Neural-network card AI for Vorarlberg Jass (Kreuz · Solo · Bodensee) — full pipeline from game engine and rule spec to MCTS-augmented behavioral cloning in TensorFlow, shipped as TensorFlow.js.

Self-play reinforcement learning AI for Agent Avenue — 17M+ games trained, C++/CUDA, evaluation suite, and playable Windows opponent.