|
I am a CS PhD Student at University of Toronto (UofT), and a Faculty Affiliate Researcher at Vector Institute, supervised by Babak Taati. Currently, I am a research intern at Samsung AI Center Toronto. Currently, my research focuses on inference-time scaling–guided generation for diffusion and flow-matching models, using targeted search, sampling, and guidance signals to steer generations toward higher quality or specific attributes. I am exploring novel guidance and steering signals (e.g., intrinsic confidence, perturbation-based methods, reward models, textual feedback, etc.) that can be used to score partial states or full trajectories. |
|
|
My research interests lie in generative modeling (flows and diffusion) for image/video/3D/4D generation and their potential to understand and construct sophisticated environments. I also have a deep interest in mathematics, optimization, and dynamic systems, particularly in their applications to generative models and machine learning. |
|
|
|
|
Javad Rajabi, Soroush Mehraban, Seyedmorteza Sadat, Babak Taati NeurIPS 2025 |
|
Vida Adeli, Ivan Klabucar, Javad Rajabi, Benjamin Filtjens, Soroush Mehraban et al. NeurIPS 2025 |
|
Soroush Mehraban, Javad Rajabi, Andrea Iaboni, Babak Taati WACV 2026 |
|
|
|
Template adapted from Jon Barron's website. |