PeopleofSuperintelligence

Researcher

Amrita Mazumdar

Amrita Mazumdar is a computer systems and graphics researcher whose work connects machine learning with video compression and interactive 3D. She studied at Columbia University and the University of Washington, then coauthored systems for low-power vision, immersive video, and neural compression. At NVIDIA Research, she coauthored QUEEN for streaming dynamic Gaussians and showed Play4D, an interactive 4D playback demo. As of October 2026, she is a senior research scientist in NVIDIA's AMRI group and coauthors Grounded-Exo2Ego, a method for generating egocentric video from exocentric footage.

Now at
NVIDIA
Works
27
Citations
224
h-index
9
Badges

Publication numbers from OpenAlex, as of 18 October 2026.

Timeline

  1. 2014–2020

    Degrees at Columbia and Washington

    She earns a BS from Columbia University in 2014, an MS from the University of Washington in 2017, and a PhD from UW in 2020.

  2. 2017

    Camera systems for vision

    In 2017, she coauthors a study of an energy-harvesting face-authentication camera and a 16-camera rig for real-time 3D-360-degree VR video.

  3. 2019

    Vignette video compression

    In 2019, she coauthors Vignette, a perception-based video-compression technique and storage manager that uses a neural network to predict saliency for transcoding.

  4. 2021

    Joining NVIDIA Research

    She joins NVIDIA Research in 2021 after founding Vignette AI, a startup focused on perception-aware video compression and storage.

  5. 2024

    Streaming dynamic Gaussians

    In 2024, she coauthors QUEEN, a framework for streaming free-viewpoint video that compresses dynamic 3D Gaussian representations.

  6. December 2025

    Showing Play4D at SIGGRAPH Asia

    In December 2025, she shows Play4D, an interactive 4D reconstruction and playback demo, at SIGGRAPH Asia Emerging Technologies.

  7. As of October 2026

    Senior research scientist at NVIDIA

    As of October 2026, she is a senior research scientist at NVIDIA Research and works in its AMRI group on 3D and 4D representations, streaming, and multimodal models. Her Grounded-Exo2Ego paper is scheduled for NeurIPS 2026 and generates egocentric video from a single exocentric video.

Career

  1. Since 2023NVIDIASenior research scientist
  2. Vignette AIFounder

Education

  1. 2020University of WashingtonPhD
  2. 2017University of WashingtonMS
  3. 2014Columbia UniversityBS

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. Principles and Techniques of Schlieren Imaging Systems2013 · 30 citations
  2. Application Codesign of Near-Data Processing for Similarity Search2018 · IEEE International Parallel and Distributed Processing Symposium · 29 citations
  3. Visual Road2019 · ACM-SIGMOD International Conference on Management of Data · 28 citations
  4. LightDB2018 · Proceedings of the VLDB Endowment · 27 citations
  5. VSS: A Storage System for Video Analytics2021 · ACM-SIGMOD International Conference on Management of Data · 24 citations

From OpenAlex, as of 18 October 2026.

  1. 14% of AI career moves in the 2020s went from a university to a companyIncludes Amrita Mazumdar. Our estimates of how university-to-company career transitions changed by decade, including roles added alongside an earlier job.
  2. Google alumni founded or led 82 other organizations, the most among the labs we compareIncludes NVIDIA. We trace 189 organizations founded or led by AI lab alumni, and 196 career moves between the labs.
  3. 44% of AI career moves go from one university to anotherIncludes NVIDIA. We estimate how 12,774 AI career moves link universities, companies, labs and government research centers, with net flows and the most common paths.
  4. Google has the highest combined h-index among AI organizationsIncludes NVIDIA. Our estimates of combined and median h-index at AI organizations, using current affiliations and OpenAlex records.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Amrita Mazumdar yet. Send us a free-licensed photo.

  1. Amrita Mazumdar | Research research.nvidia.com. Checked 18 October 2026.
  2. [1706.03864] Exploring Computation-Communication Tradeoffs in Camera Systems arxiv.org. Checked 18 October 2026.
  3. [1902.01372] Vignette: Perceptual Compression for Video Storage and Processing Systems arxiv.org. Checked 18 October 2026.
  4. Amrita Mazumdar - home amritamaz.net. Checked 18 October 2026.
  5. [2412.04469] QUEEN: QUantized Efficient ENcoding of Dynamic Gaussians for Streaming Free-viewpoint Videos arxiv.org. Checked 18 October 2026.
  6. Grounded-Exo2Ego — NVIDIA research.nvidia.com. Checked 18 October 2026.