PeopleofSuperintelligence

Researcher

Adrien Bardes

Adrien Bardes is a computer vision researcher whose papers apply self-supervised learning to images, motion and video. He earned a master’s in Mathematics, Vision and Learning and bachelor’s degrees in theoretical computer science and mathematics and computer science, began a PhD with Inria’s WILLOW team in 2020, and coauthored VICReg in 2021. He later coauthored VICRegL and MC-JEPA, and his 2023 paper linking contrastive and covariance based non-contrastive criteria received an ICLR Honorable Mention. In April 2024, he completed a PhD at ENS Paris on joint-embedding predictive architectures; in 2025, V-JEPA 2 used more than one million hours of internet video for understanding and prediction and enabled zero-shot robot planning. As of October 2026, he is affiliated with AMI Labs; in 2026, he coauthored MJEPA for audio-visual learning and spoke at an ECCV workshop about dense and multimodal JEPA representations.

Now at
AMI Labs
Links
Website, Google Scholar, X, LinkedIn and GitHub

Timeline

  1. 2020–2021

    Starting a PhD and VICReg

    His degrees include a master’s in Mathematics, Vision and Learning and bachelor’s degrees in theoretical computer science and mathematics and computer science. He begins a PhD with Inria’s WILLOW team in October 2020 under Jean Ponce. In 2021, he coauthors VICReg, which explicitly avoids representation collapse with variance and covariance regularization.

  2. 2022

    VICRegL for local features

    In 2022, he coauthors VICRegL with Jean Ponce and Yann LeCun. The method learns global and local image features for classification and segmentation.

  3. 2023

    An ICLR honorable mention

    In 2023, he coauthors “On the duality between contrastive and non-contrastive self-supervised learning,” which relates contrastive and covariance based criteria algebraically. The paper receives an ICLR 2023 Honorable Mention.

  4. 2023

    MC-JEPA for motion features

    In 2023, he coauthors MC-JEPA with Jean Ponce and Yann LeCun, a self-supervised architecture that learns optical-flow and content features within one shared encoder.

  5. 2024

    A PhD on JEPA

    In April 2024, he completes a PhD at ENS Paris with a thesis on self-supervised learning using joint-embedding predictive architectures, advised by Jean Ponce and Yann LeCun. His PhD role at Meta AI ends in April 2024.

  6. 2025

    V-JEPA 2 and robot planning

    In 2025, he coauthors V-JEPA 2, pretrained on a video and image dataset comprising over one million hours of internet video for understanding and prediction. Its action-conditioned version enables zero-shot robot planning and picking and placing objects with Franka arms.

  7. June 2026

    MJEPA for audio-visual learning

    In June 2026, he coauthors MJEPA, a joint-embedding predictive architecture for audio-visual learning. It uses one shared encoder and a predictive objective applied within and across modalities.

  8. As of October 2026

    An AMI Labs JEPA talk

    At the September 2026 ECCV workshop, he speaks for AMI Labs on learning dense and multimodal representations with JEPA. As of October 2026, he is affiliated with AMI Labs.

Career

  1. NowAMI Labs
  2. Until 2024MetaPhD
  3. From 2020InriaPhD researcher

Education

  1. 2024École Normale SupérieurePhD
  2. From 2020InriaPhD

From the biography.

  1. Language models reach 33% of 2020s entrants, and company shares run from 8% in pattern recognition to 91% in AI infrastructureIncludes Adrien Bardes. We estimate how research topics in AI changed across seven entry decades (6,853 people from the 1960s to the 2020s) and how 60 major topics split between companies (8% to 91%) and universities (10% to 93%).
  2. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes AMI Labs. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes AMI Labs. 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. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Meta. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.

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 Adrien Bardes yet. Send us a free-licensed photo.

  1. Adrien Bardes adrien987k.github.io. Checked 21 October 2026.
  2. WILLOW - 2022 - Rapport annuel d'activité radar.inria.fr. Checked 21 October 2026.
  3. VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning arxiv.org. Checked 21 October 2026.
  4. VICRegL: Self-Supervised Learning of Local Visual Features arxiv.org. Checked 21 October 2026.
  5. ICLR 2023 Awards iclr.cc. Checked 21 October 2026.
  6. MC-JEPA: A Joint-Embedding Predictive Architecture for Self-Supervised Learning of Motion and Content Features arxiv.org. Checked 21 October 2026.
  7. WILLOW - 2024 - Rapport annuel d'activité radar.inria.fr. Checked 21 October 2026.
  8. V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning arxiv.org. Checked 21 October 2026.
  9. MJEPA: A Simple and Scalable Joint-Embedding Predictive Architecture for Audio-Visual Learning arxiv.org. Checked 21 October 2026.
  10. How to Build World Models|ECCV'26 eccv26wmeai.github.io. Checked 21 October 2026.