PeopleofSuperintelligence

Researcher

Mehdi Cherti

Mehdi Cherti is a machine-learning researcher whose work includes open image-text datasets, vision-language model scaling, and multimodal video data. He earned a PhD in Machine Learning from Université Paris Saclay in 2018 after research on deep generative models and novelty generation. In 2022, he coauthored LAION-5B, an open dataset of 5.85 billion image-text pairs that received a NeurIPS Outstanding Paper Award; in 2023, he coauthored the DataComp benchmark and a CVPR study on CLIP scaling. His 2025 work compares CLIP and MaMMUT with scaling laws, and in August 2026 he coauthored LAION-BVD, an open video dataset for multimodal pretraining with 10 million hours of video. As of October 2026, he is a postdoctoral researcher at Jülich Supercomputing Center and co-founder of LAION; he is also a founding member and senior core researcher in JSC's SLAMPAI Lab.

Now at
Jülich Supercomputing Centre and LAION
Badges

Timeline

  1. 2017–2018

    A PhD in machine learning

    By December 2017, his PhD research explores deep generative models and how to use them to study and experiment with novelty generation. He earns a PhD in Machine Learning from Université Paris Saclay in 2018.

  2. 2022

    LAION-5B earns an award

    In 2022, he coauthors LAION-5B, an open dataset of 5.85 billion CLIP-filtered image-text pairs, including 2.32 billion English pairs. NeurIPS names LAION-5B an Outstanding Paper in its 2022 awards.

  3. 2023

    The DataComp benchmark

    In 2023, he coauthors DataComp, a benchmark built around a candidate pool of 12.8 billion image-text pairs from Common Crawl. The benchmark lets participants compare data-filtering methods using standardized CLIP training and 38 downstream test sets.

  4. 2023

    Scaling CLIP with OpenCLIP

    In 2023, he coauthors the CVPR paper Reproducible scaling laws for contrastive language-image learning. It studies CLIP pretraining with the public LAION dataset and the open-source OpenCLIP repository.

  5. 2025

    Comparing CLIP and MaMMUT

    In 2025, he coauthors a NeurIPS paper that uses scaling laws to compare CLIP and MaMMUT across the DataComp, DFN and Re-LAION datasets. The study presents evidence that MaMMUT improves more with scale and uses samples more efficiently than standard CLIP.

  6. August 2026

    LAION-BVD video dataset

    In August 2026, he coauthors LAION-BVD, an open video dataset for multimodal pretraining. It contains 1.3 billion video URLs and 80 million videos with a total duration of 10 million hours.

  7. As of October 2026

    JSC and LAION research roles

    As of October 2026, he is a postdoctoral researcher at Jülich Supercomputing Center and co-founder of LAION. He is a founding member and senior core researcher in the Scalable Learning and Multi-Purpose AI (SLAMPAI) Lab at JSC.

Career

  1. NowJülich Supercomputing CentrePostdoctoral Researcher
  2. NowLAIONCo-founder

Education

  1. 2018Université Paris-SaclayPhD in Machine Learning

From the biography.

  1. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Mehdi Cherti. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  2. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Université Paris-Saclay. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  3. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Université Paris-Saclay. 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.
  4. Google has the highest combined h-index among AI organizationsIncludes Université Paris-Saclay. 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.

Sources and credits

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

  1. Mehdi Cherti – First post mehdidc.gitlab.io. Checked 21 October 2026.
  2. Mehdi Cherti - Postdoc Researcher @ Jülich / LAION mehdidc.github.io. Checked 21 October 2026.
  3. NeurIPS 2022 Awards neurips.cc. Checked 21 October 2026.
  4. LAION-5B: An open large-scale dataset for training next generation image-text models papers.nips.cc. Checked 21 October 2026.
  5. DataComp: In search of the next generation of multimodal datasets arxiv.org. Checked 21 October 2026.
  6. Reproducible scaling laws for contrastive language-image learning arxiv.org. Checked 21 October 2026.
  7. Scaling Laws for Robust Comparison of Open Foundation Language-Vision Models and Datasets arxiv.org. Checked 21 October 2026.
  8. LAION-BVD: A 10-Million-Hour Open Video Dataset for Multimodal Pre-training arxiv.org. Checked 21 October 2026.
  9. Team | LAION laion.ai. Checked 21 October 2026.