Rishit Dagli

CS, Math Undergrad at UofT

rishit.png rishit.png

I am very interested in learning algorithms, computer vision, graphics, learning theory, and math (number theory and topology).

I am currently on a break from my undergrad and I am working at NVIDIA on the intersection of AI, vision, and graphics research. After I switched boats to research I interned at Qualcomm AI Research in 2024 with Roland Memisevic and Guillaume Berger and at Civo in 2023 with Josh Mesout.

In a past life, I used to work on software engineering and robotics. I used to contribute extensively to/ maintain some popular open-source projects which can be found on my github and software.

news

Oct 3, 2024 We released a new VLM and large-scale dataset for video understanding. Arxiv. Dataset. (code release soon, in the hands of corporate overlords)
Jun 18, 2023 We released the first vision (images and video)-spatial audio model as a step towards complete generation. Arxiv. Code and Web Demo.

selected publications

  1. squeeze3d.png
    Squeeze3D: Your 3D Generation Model is Secretly an Extreme Neural Compressor
    Rishit Dagli, Yushi Guan, Sankeerth Durvasula, and 2 more authors
    arXiv 2025
    Project Webpage Abs arXiv PDF Code Dataset
  2. qivd.png
    Can Vision-Language Models Answer Face to Face Questions in the Real-World?
    Reza Pourreza*, Rishit Dagli*, Apratim Bhattacharyya, and 3 more authors
    arXiv 2025 (* joint first authors)
    Project Webpage Abs arXiv PDF CVPR Challenge
  3. s2s.png
    SEE-2-SOUND: Zero-Shot Spatial Environment-to-Spatial Sound
    Rishit Dagli, Shivesh Prakash, Robert Wu, and 1 more author
    SIGGRAPH Posters 2025
    Project Webpage Abs arXiv PDF Code
  4. nerfus.png
    NeRF-US: Removing Ultrasound Imaging Artifacts from Neural Radiance Fields in the Wild
    Rishit Dagli, Atsuhiro Hibi, Rahul G. Krishnan, and 1 more author
    PMLR 2024
    Project Webpage Abs arXiv PDF Code Dataset