I am a PhD student in Computer Science and Engineering at The Ohio State University, specializing in interpretability and fairness in machine learning. I hold a Bachelor's degree in Computer Science from Rutgers, The State University of New Jersey.
Previously, I had the opportunity to work with Prof. Eshed Ohn-Bar on advancing vehicle keypoint detection under extreme occlusion and truncation. This work contributed to improving fine-grained perception in autonomous driving by leveraging novel geometric loss functions and transformer-based architectures, significantly boosting detection accuracy in challenging real-world scenarios.
My research focuses on fair and interpretable AI, aiming to understand the internal representations and decision-making processes of complex models. My current work centers on high-order attention mechanisms for transformers, aiming to enhance their interpretability and reasoning capabilities in multimodal and structured data settings.
Here are some of my recent projects:
Examining the Reversal Curse on Logical Equivalence (Fall 2024)
The Reversal Curse paper highlights a simple task that these models fail at. If the model has seen A is B, it is not guaranteed that the model can generalize B is A - this is coined as Reversal Curse in the paper. Expanding on this, my work investigates whether this limitation extends to logical equivalences—specifically, whether models trained on "A implies B" can infer "not B implies not A." This research evaluates the model’s ability to reason beyond memorization.