I'm a Msc in Applied Computing student at UofT with an interest in developing decision making agents for difficult environments.
- General Game Playing Agent: This is a new project that is building off what I've learned from 2048 AI. This time, I am designing the agent to play any game that can be modelled as an MDP, without encoding the rules or rewards ahead of time. For now I am building and testing against the performance of the 2048 bot I made previously, but in the future I plan to extend to multiplayer games and games with large action spaces.
- Borderline Project: I'm working with Jessica Thompson and team from the University of Waterloo to explore the relationship between sound and socioeconomic indicators in urban environments. My contribution to the project is through the app, where I have improved the scalability and reliability of serving Canada-wide maps and datasets. Have a look on IOS: Borderline App
- 2048 AI: I'm comparing traditional artificial intelligence techniques to maximize the performance of an AI agent. I've reached the 2048 tile in 96% of games by combining expectimax with a BayesOpt tuned heuristic that rewards strategies arranges the tiles in a monotonic snake. See if you can beat me!