About

I’m currently an Applied Scientist at Amazon AWS Fundamental Research Team, working on LLMs and agents. Before that, I was a PhD student in Data Science at New York University advised by Yann LeCun and Kyunghyun Cho. During my PhD, I was also a visiting researcher at Meta AI Research from October 2022 to October 2024. I’m broadly interested in representation learning, world models, planning, reinforcement learning, and autonomous vehicles.

Before, I completed my undergrad at the University of Warsaw in Poland. I also have extensive industry experience from doing internships at at Google, Nvidia and Jane Street.

Publications

Ordered chronologically:

1. Learning from Reward-Free Offline Data: A Case for Planning with Latent Dynamics Models
   Vlad Sobal, Wancong Zhang, Kyunghyun Cho, Randall Balestriero, Tim G. J. Rudner, Yann LeCun
   (Conference on Neural Information Processing Systems (NeurIPS) 2025)
   (Best Paper Award at ICML 2025 Building Physically Plausible World Models Workshop)

2. X-Sample Contrastive Loss: Improving Contrastive Learning with Sample Similarity Graphs
   Vlad Sobal, Mark Ibrahim, Randall Balestriero, Vivien Cabannes, Diane Bouchacourt, Pietro Astolfi, Kyunghyun Cho, Yann LeCun
   (International Conference on Learning Representations (ICLR) 2025)

3. Hierarchical World Models as Visual Whole-Body Humanoid Controllers
   Nicklas Hansen, Jyothir S V, Vlad Sobal, Yann LeCun, Xiaolong Wang, Hao Su
   (International Conference on Learning Representations (ICLR) 2025)

4. Gradient-based Planning with World Models
   Jyothir S V, Siddhartha Jalagam, Yann LeCun, Vlad Sobal
   (Generative Models for Decision Making Workshop at ICLR 2024)

5. A Cookbook of Self-Supervised Learning
   Joint work with many other collaborators, contributed the section on RL

6. Light-weight probing of unsupervised representations for Reinforcement Learning
   Wancong Zhang, Anthony GX-Chen, Vlad Sobal, Yann LeCun, Nicolas Carion’ (Reinforcement Learning Conference 2024)

7. Joint Embedding Predictive Architectures Focus on Slow Features
   Vlad Sobal, Jyothir S V, Siddhartha Jalagam, Nicolas Carion, Kyunghyun Cho, Yann LeCun
   (SSL Theory and Practice Workshop at NeurIPS 2022)

8. Separating the World and Ego Models for Self-Driving
   Vlad Sobal, Alfredo Canziani, Nicolas Carion, Kyunghyun Cho, Yann LeCun
   (Generalizable Policy Learning in the Physical World Workshop at ICLR 2022)

9. TiCo: Transformation Invariance and Covariance Contrast for Self-Supervised Visual Representation Learning
   Jiachen Zhu, Rafael M. Moraes, Serkan Karakulak, Vlad Sobal, Alfredo Canziani, Yann LeCun