Vijay Gohil

I’m a machine learning and computer vision engineer at 2K Games and a visiting researcher at the NYU AI4CE Lab. My work sits at the intersection of production ML systems and research, with a focus on 3D perception, multiview reasoning, and vision interpretability.

My research interests include computer vision, vision transformer interpretability, and multi-camera 3D reconstruction. I’m particularly interested in mechanistic interpretability for vision models: understanding what vision transformers represent internally and how those representations can be measured, compared, and improved.

On the applied side, I work on object detection, pose estimation, 3D triangulation, and animation-adjacent ML tooling for video games, where geometric understanding has to meet production reliability and real-time constraints.