Mido Assran
Scientist at Advanced Machine Intelligence (AMI); first author of V-JEPA 2 and I-JEPA from Meta FAIR
- Scientist and Member of Technical Staff, founding team, Advanced Machine Intelligence (AMI)1
Mahmoud (Mido) Assran is a scientist and member of technical staff on the founding team of Advanced Machine Intelligence (AMI), and was previously a research scientist in Meta's Fundamental AI Research (FAIR) lab. Assran works on self-supervised learning and world models built on Joint-Embedding Predictive Architectures (JEPA), and is first author of I-JEPA (2023) and of V-JEPA 2 (2025).1235
Career
Assran completed a PhD in electrical and computer engineering at McGill University and Mila, supported by a Vanier Scholarship, with earlier work on parallelizing deep reinforcement learning and large-scale numerical optimization. Assran then worked as a research scientist in Meta FAIR on self-supervised learning from video for reasoning and planning in the physical world.21
Assran's site now describes Assran as a scientist and member of technical staff on the founding team of Advanced Machine Intelligence (AMI). At Meta, Assran was also program lead of the Meta AI Mentorship Program for doctoral students.1
JEPA research
Assran was first author of I-JEPA (January 2023), which learns image representations by predicting in latent space, and co-authored V-JEPA (February 2024), with Adrien Bardes as first author. Assran also co-authored DINOv2 and work on learning and leveraging world models in visual representation learning (ICML 2024).341
Assran is first author of V-JEPA 2 (June 2025), on self-supervised video models for understanding, prediction and planning, and a co-author of V-JEPA 2.1 (March 2026), which targets dense features in video self-supervised learning.56
Work
V-JEPA 2 ViT-g/16 (384 px)Open weights
First author of "V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning" (2025), the paper behind the V-JEPA 2 model family.521
V-JEPA 2.1 ViT-G/16 (384 px)Open weights
Co-author of "V-JEPA 2.1: Unlocking Dense Features in Video Self-Supervised Learning" (2026), which follows V-JEPA 2.6
I-JEPA (external site: arxiv.org)Published research
First author of "Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture" (CVPR 2023).31
Career timeline
Jan 2023
Posted I-JEPA, joint-embedding predictive self-supervised learning from images, as first author.3
Feb 2024
Co-authored V-JEPA, revisiting feature prediction for visual representations from video.4
Jun 2025
Posted V-JEPA 2 as first author.5
Mar 2026
Co-authored V-JEPA 2.1 on dense features in video self-supervised learning.6
Other documented roles
Roles the sources document but that are not confirmed as current.
Honors and recognition
- Vanier Scholarship · Vanier Canada Graduate Scholarships program1
Related people
More sourced profiles at the same organization, in the same field, or from the same school.
- Abdelrahman MohamedAlso at Facebook AI Research
- Yann LeCunAlso at Meta
- Adina WilliamsAlso at Meta
- Amnon ShashuaComputer vision
- Aditya GroverAlso at Meta Fundamental AI Research
- Byron BootsRobotics
- Ahmad Al-DahleAlso at Meta
- Deepak PathakRobotics
- Alban DesmaisonAlso at Meta
- Dieter FoxRobotics
- Albert JiangAlso at FAIR
- Geoffrey HintonMachine-learning theory
Is this you?
Profiles cover public professional work only, and every line cites a source. If something is out of date, missing or wrong, send a correction with a link to an official page and we will update it. Suggest a correction
Add this badge to your website or GitHub profile README. It links back to this page. It is not an endorsement or a rating.
Markdown (for a README)
[](https://unitedstatesofamericasuperintelligence.com/people/mido-assran/)
HTML
<a href="https://unitedstatesofamericasuperintelligence.com/people/mido-assran/"><img src="https://unitedstatesofamericasuperintelligence.com/badges/profile.svg" alt="Mido Assran — Profile on USASI" height="20"></a>
This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.