Portfolio Project

Nonogram Solver

Reinforcement Learning (RL)

Data Science Machine Learning Python PyTorch AWS Docker

Demo

STAR Summary

Situation
I wanted to see if an RL agent could learn to solve Nonogram puzzles.
Task
Owned the end-to-end build, from implementation through the final deliverable.
Action
  • Generated large batches of 5x5 puzzles and trained a hybrid CNN + Transformer policy network to solve them.
  • Shaped rewards around unique guesses, row/column completions, and full-board solves to guide exploration.
Result
  • Achieved strong solve rates on held-out 5x5 boards (see GitHub for evaluation details).

Notes

The solver focuses on generated 5x5 puzzles; scaling to larger puzzles would need curriculum learning or search.