Rohit Jena was a Ph.D. student in Computer and Information Science at the University of Pennsylvania, advised by Pratik Chaudhari and Jim Gee. Rohit’s research focused on correspondence matching problems such as image registration: how to build task-specific invariances into these problems, which self-supervised representations help discover them, and how well those representations transfer.
At PICSL, Rohit developed FireANTs, a fast, training-free, GPU-accelerated registration algorithm published in Nature Communications. Rohit also wrote “Deep Learning in Medical Image Registration: Magic or Mirage?” (NeurIPS 2024), which examines when classical optimization and learning-based registration methods each work best, and offers a recipe for choosing between them.
During the Ph.D., Rohit interned with the NeMo team at NVIDIA, working on aligning text-to-image diffusion models, and at Amazon Lab126, working on mesh-NeRF hybrids for rigged 3D avatars. Before Penn, Rohit earned a master’s in Robotics at Carnegie Mellon University and a bachelor’s in Computer Science and Engineering at the Indian Institute of Technology Bombay.