Projects

• GenRL CODE|DOCS
  A PyTorch reinforcement learning library centered around reproducible and generalizable algorithm implementations. Currently contributing to the Model Based part of the library

• Exploration of RL algorithms

• Reinforcement Learning on Easy21 CODE
  Developed an environment to simulate playing of Easy21 and tried out Monte Carlo Control and TD learning to learn the value functions

• Deep Q learning on Atari Games CODE
  Experimented with DQN algorithms to play Ms Pacman in a simulated gym environment

• Study of optimizers CODE
  Study and comparison of various popular optimizers, coded from scratch using JAX

• One shot Classification using transfer learning CODE
  Used transfer learning techniques to improve performance of a Siamese network for one shot learning on the Omniglot dataset.