About         Academic Research         Experience        Contact

Overview

Starting early 2015, I have pursued state-of-the-art interesting research problems on applied and pure machine learning. The problems have been inspired from three major domains - Natural Language, Healthcare(2017-2019) at Harvard University and Quant Finance(2015-2017) at the Execution Algorithms group at Goldman Sachs. My primary research under Prof. Finale Doshi-Velez focusses on semi-supervised learning, generative modeling and Reinforcement based imitation learning(inverse reinforcement learning) in healthcare, with current focus on sepsis treatment in ICUs. My other parallel research project advised by Prof.Alexander Rush as part of CS287 at Harvard, focusses on variational and adversarial generative deep models for automated natural language generation. Testing links here

Projects at Harvard

2018

1. Inverse Reinforcement/Imitation Learning for Sepsis Treatment
2. Variational and Adversarial Deep Generative Models for Natural Language
3. Input Convex Neural Networks
4. Neural Fictitious Self Play Bot for Poker(Deep Learning/RL)
5. Genetic Algorithm based hyperparameter tuning for deep networks
6. Minimax Beam Search Agent for Connect-6 Game

Projects(Goldman Sachs)