PeopleofSuperintelligence

Researcher

Yuval Atzmon

Yuval Atzmon is a researcher at NVIDIA Research whose coauthored papers on personalizing text-to-image models include Perfusion. After B.Sc. and M.Sc. degrees in electrical engineering at the Technion, he earned a Ph.D. under Gal Chechik at Bar-Ilan University, and his early papers from 2016 to 2019 dealt with generalization to new compositions and generalized zero-shot learning. In February 2020, he joined NVIDIA Research as a research scientist, and his causal approach to compositional zero-shot recognition became a NeurIPS 2020 spotlight. Textual Inversion (ICLR 2023), Perfusion (SIGGRAPH 2023) and ConsiStory (SIGGRAPH 2024) followed, each coauthored with Rinon Gal and Gal Chechik. He is a member of NVIDIA Research's Tel-Aviv lab, where he works on controllable and personalized image and video generation.

Now at
NVIDIA
Works
39
Citations
1,537
h-index
15
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2016

    Generalizing to new compositions

    His B.Sc. and M.Sc. in electrical engineering come from the Technion, and he does his Ph.D. under Gal Chechik at Bar-Ilan University. In 2016, he coauthors a paper with Jonathan Berant, Vahid Kezami, Amir Globerson and Chechik in which a structured prediction model generalizes to new compositions of known entities with about seven times the accuracy of an LSTM model.

  2. 2019

    Adaptive confidence smoothing

    In 2019, he and Gal Chechik present COSMO, a method for generalized zero-shot learning, at CVPR. It splits the task among a gating classifier, a zero-shot expert and a seen-class expert, and it outperforms earlier models on four standard benchmarks.

  3. 2020

    NVIDIA and a NeurIPS spotlight

    He joins NVIDIA Research as a research scientist in February 2020. In 2020, he also coauthors A causal view of compositional zero-shot recognition with Felix Kreuk, Uri Shalit and Gal Chechik; the paper is accepted to NeurIPS as a spotlight and frames zero-shot inference as finding which intervention caused the image.

  4. 2022–2023

    Textual Inversion at ICLR

    In 2022, he coauthors An Image is Worth One Word with Rinon Gal, Yuval Alaluf, Or Patashnik, Amit Bermano, Gal Chechik and Daniel Cohen-Or. Textual Inversion learns a new word for a user-provided concept from 3 to 5 images in the embedding space of a frozen text-to-image model, and the paper appears at ICLR in 2023 as a spotlight.

  5. 2023

    Perfusion at SIGGRAPH

    In 2023, he coauthors Perfusion with Yoad Tewel, Rinon Gal and Gal Chechik. The text-to-image personalization method uses dynamic rank-1 updates to the underlying model and locks new concepts' cross-attention Keys to their superordinate category to avoid overfitting; the paper is accepted to SIGGRAPH 2023.

  6. 2024

    ConsiStory and an NVIDIA blog post

    In February 2024, he coauthors the NVIDIA Technical Blog post Generative AI Research Spotlight: Personalizing Text-to-Image Models with Gal Chechik and Rinon Gal, covering Perfusion and the encoder-for-tuning method E4T. He is also a coauthor of ConsiStory, a training-free approach to consistent text-to-image generation accepted to the journal track of SIGGRAPH 2024.

  7. 2025

    Identity and motion in video

    In 2025, he publishes Identity-Motion Trade-offs in Text-to-Video Generation at BMVC with Rinon Gal, Yoad Tewel, Yoni Kasten and Gal Chechik, all at NVIDIA Research. The paper finds that self-attention query features in text-to-video diffusion models affect both layout and subject identity, and it uses query feature injection for zero-shot motion transfer and consistent multi-shot video generation.

  8. As of October 2026

    Member of NVIDIA's Tel-Aviv lab

    As of October 2026, he is a member of NVIDIA Research's Tel-Aviv lab, which Gal Chechik leads. His work on generative vision models spans video, natural language and action, with a focus on controllable and personalized image and video generation.

Career

  1. Since 2020NVIDIAResearch Scientist

Education

  1. TechnionB.Sc.
  2. Bar-Ilan UniversityPh.D.

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

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Yuval Atzmon. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  2. Google has the highest combined h-index among AI organizationsIncludes NVIDIA. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes NVIDIA. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes NVIDIA. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

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

  1. Yuval Atzmon | Research research.nvidia.com. Checked 20 October 2026.
  2. Learning to generalize to new compositions in image understanding arxiv.org. Checked 20 October 2026.
  3. Adaptive Confidence Smoothing for Generalized Zero-Shot Learning arxiv.org. Checked 20 October 2026.
  4. Author: Yuval Atzmon | NVIDIA Technical Blog developer.nvidia.com. Checked 20 October 2026.
  5. A causal view of compositional zero-shot recognition arxiv.org. Checked 20 October 2026.
  6. An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversion | NVIDIA Research Tel-Aviv Lab research.nvidia.com. Checked 20 October 2026.
  7. Key-Locked Rank One Editing for Text-to-Image Personalization arxiv.org. Checked 20 October 2026.
  8. Training-Free Consistent Text-to-Image Generation arxiv.org. Checked 20 October 2026.
  9. Generative AI Research Spotlight: Personalizing Text-to-Image Models | NVIDIA Technical Blog developer.nvidia.com. Checked 20 October 2026.
  10. Identity-Motion Trade-offs in Text-to-Video Generation | BMVC 2025 bmvc2025.bmva.org. Checked 20 October 2026.
  11. NVIDIA Research Tel-Aviv Lab research.nvidia.com. Checked 20 October 2026.