Alok Shah

I'm a student at University of Pennsylvania where I study Computer Science (B.S.E), Mathematics (B.A.), and Electrical Engineering (M.S.E) with a focus on Optimization and Control.

Outside of class I lead MLR@Penn, the undergraduate AI and ML research organization and community at Penn and serve on the Student Advisory Board for the Wharton AI & Analytics Initiative.

In my free time, I enjoy history and geography trivia, traveling, trying new food trucks and restaurants around Philly, and burning off those calories by playing pickup soccer.

I'm open to collaborations or talking about anything related to ML and research -- I especially enjoy mentoring younger students interested in getting involved in the field!

Email  /  CV  /  Twitter  /  Github

Research

I'm broadly interested in playing simple tricks on models to improve empirical and theoretical bounds alike on performance across modalities. Currently, I'm developing methods to induce more robust, explainable, and generalizable reasoning capabilities in generative models while ensuring their responsible use in practice.

Explored how to coerce communication between token-level checkpoints for more interpretable, capable models

Developed in-context optimization method leveraging weak learners to improve reasoning in strong large language models without additional finetuning

Integrating conformal prediction into offline reinforcement learning, providing statistically robust uncertainty quantification while addressing overestimation bias in Q-value functions

ESE 5390 Final Project: A unified framework for model compression and adversarial robustness by exploiting the geometry of class boundaries

Investigated the efficacy of using large language models as surrogate reward functions for reinforcement learning in the ALFWorld environment.