>_ Game Arcade
browser games built with hand-coded JS · RL agents in training
← back to portfolio
// playable now
[::]
escape_grid.exe
4 participants vs 1 hunter on a 4×4 grid. Unlock doors, dodge traps, reach the exit. Hunter places traps before the game starts.
turn-based 4×4 grid traps auto-play
[⬡]
gomoku.exe
Classic 5-in-a-row strategy game on a 15×15 board. Play against a minimax AI at three difficulty levels. RL agent slot reserved for future training.
strategy 15×15 minimax AI RL soon

// in the lab
[RL]
gomoku_rl_agent.py
Train a policy gradient agent to play Gomoku. Export weights to JSON, load in browser. Visualise learned patterns — double-threes, forks, diagonal threats.
reinforcement learning PyTorch explainability