Soroush Mahdi

I recently graduated from Amirkabir University of Technology (Tehran Polytechnic) with an MSc in AI, where I successfully defended my thesis on improving the robustness of deep neural networks against adversarial attacks, under the guidance of Prof. Amir Mazlaghani. During this time, I also served as a research assistant in the Statistical Data Analysis Laboratory, focusing on adversarial robustness in deep learning.

I am currently a research assistant in the Autonomous & Intelligent Systems Lab (AISL) at Amirkabir University of Technology, where I work on 3D computer vision, with a particular emphasis on structure-from-motion (SfM) techniques powered by deep learning.

My research interests include:

• 3D Computer Vision
• Computer Vision
• Computational Neuroscience
• Understanding the inner workings of Deep Learning models
• Generative AI

I’m passionate about exploring the intersection between 3D vision, brain-inspired modeling, and generative techniques, and I aim to pursue a Ph.D. to further contribute to these areas.

If you’d like to collaborate on Deep Learning projects, explore computational neuroscience, or discuss Ph.D. opportunities, please feel free to reach out—I’d love to connect!