Hyperphantasia: A Benchmark for Evaluating the Mental Visualization Capabilities of Multimodal LLMs
Bio
I am a PhD student in Electrical and Computer Engineering at the University of Southern California, where I work with Prof. Mahdi Soltanolkotabi at the USC Center on AI Foundations for Science (AIF4S). My research focuses on reasoning and reliability in generative models, particularly vision-language and diffusion models.
I received my M.Sc. in Electrical and Computer Engineering from USC in 2025 and dual B.S. degrees in Electrical Engineering and Computer Science in 2022 from Sharif University of Technology, where I worked on detection algorithms for computer vision.
Education
PhD Electrical Engineering
University of Southern California
MSc Electrical Engineering
University of Southern California
BSc Computer Science
Sharif University of Technology
BSc Electrical Engineering
Sharif University of Technology
Interests
Large Vision Language Models
Reasoning
Reinforcement Learning
Time Series
📚 My Research
My research focuses on advancing reasoning in generative models, particularly vision-language models and diffusion models. I am interested in moving beyond pattern recognition toward structured and explainable reasoning. Recent projects include probing mental visualization capabilities, optimizing prompts for diffusion models, and evaluating the reliability of multimodal medical foundation models, with the broader goal of developing more trustworthy and generalizable AI systems.
Feel free to reach out to collaborate. 🤝
Featured Publications
CryptoMamba: Leveraging State Space Models for Accurate Bitcoin Price Prediction
MediConfusion: Can you trust your AI radiologist? Probing the reliability of multimodal medical foundation models
Recent Publications
(2025).
Hyperphantasia: A Benchmark for Evaluating the Mental Visualization Capabilities of Multimodal LLMs.
NeurIPS.
(2025).
LANTERN: A Machine Learning Framework for Lipid Nanoparticle Transfection Efficiency Prediction.