I’m a PhD student in Computer Science at Dalhousie University, supervised by Prof. Janarthanan Rajendran. My research is on meta-reinforcement learning: I work on how agents can learn intrinsic reward functions that make downstream learning more sample-efficient.

Background

I completed my BSc in Engineering at the University of Moratuwa, specializing in Computer Science and Engineering. After graduating I stayed on for some time as a contract lecturer before moving to Dal for my PhD.

Before settling on RL, I worked across a range of applied ML problems: scene-text recognition, few-shot multispectral segmentation, and ML-driven sample-site selection for Mars-analog field studies. These are reflected in my publications and the blog, which is where most of my expository writing lives.

Research interests

  • Reinforcement learning. Meta-RL, intrinsic motivation, reward shaping, sample efficiency.
  • Computer vision. Segmentation under data-scarce regimes, representation learning.
  • The math underneath. Probabilistic methods, optimization, the theory that connects all of the above.

Elsewhere

The fastest way to reach me is by email. You can also find me on GitHub and Twitter.