Exploring combinatorial games and search algorithms where intelligent agents make optimal decisions within perfect information environments. Projects experiment with minimax, alpha-beta pruning, transposition tables, and novel memory structures such as MemoryMap to test the boundaries between search, intuition and artificial intelligence.
Featured Work
BitSnake: Competitive Snake with Intelligent Pathfinding
BitSnake broke away from turn-based board games, introducing multiple agents, simultaneous decisions, and elements of randomness. It marked the leap into stochastic, imperfect-information environments, transforming simple pathfinding agents into competitors in a living ecosystem—an experiment in both survival and strategy.
PythonFlaskWebSocketsBFS+1 more
Strategic Gomoku AI: Bayesian Approximation Using Memory-Mapped Structure
Gomoku pushed beyond the limits of traditional search, requiring adaptive mechanisms like the MemoryMap to bias decisions with accumulated experience. Inspired by AlphaGo, this project represents the step from classical algorithms toward systems that evolve with history, embodying the transition from deterministic certainty to adaptive intuition.
Javaminimaxalpha-betatransposition tables
Connect Four AI: Classic Adversarial Agent
Building on Tic-Tac-Toe, Connect Four expanded the state space and demanded techniques like Alpha-Beta pruning, heuristics, and bitboard representations. It demonstrated how efficiency and selective search could turn brute-force play into emergent strategy, bridging the gap toward more complex games like Gomoku.
Pythonminimaxalpha-beta
TicTacToe AI: Human vs AI Baseline
Tic-Tac-Toe introduced the principles of adversarial search in a perfectly deterministic, zero-sum setting. This project marks the starting point of the journey, showing how simple recursion can capture the essence of strategy and set the stage for deeper explorations.
Pythonminimaxalpha-beta