Research Assistant · RIML Lab · M.Sc. Student
Compositional Visual Generation · Diffusion Models · Visual Reasoning
Download CV
I'm an M.Sc. student in Computer Engineering at Sharif University of Technology and a Research Assistant at the RIML Lab, where I work on compositional visual generation, text-to-image diffusion models, and cross-modal reasoning.
My research focuses on teaching machines to combine visual elements in structured and creative ways — spanning generative modeling, evaluation, and semantic alignment across vision and language. I'm always open to connecting with others interested in machine learning, computer vision, and AI research.
Publications
Certificates
Skills
Experiences
Transactions on Machine Learning Research (TMLR) 2026
A reward-guided optimization framework for initial noise in diffusion models, combining discrete exploration and continuous refinement for compositional alignment.
NeurIPS 2025 Workshop: Generative and Protective AI for Content Creation (GenProCC)
A correlation study of compositional text-to-image metrics, benchmarking category-specific and cross-category agreement with human evaluations.
NeurIPS 2025 Workshop: Generative and Protective AI for Content Creation (GenProCC)
A taxonomy of attribute, relation, and object hallucinations in text-to-image models, framing hallucination as an upper bound for richer evaluation.
ICLR 2026 Workshop: From Human Cognition to AI Reasoning (HCAIR)
RebusBench is a benchmark of 1,164 rebus puzzles for evaluating abstract cognitive visual reasoning in LVLMs.
CVPR 2026 Workshop: Machine Unlearning for Computer Vision (MUC)
Evaluating unlearned diffusion models through the lens of compositionality.
Under review at ICLR 2027
An adaptive, concept-conditioned noise sampling strategy that improves robust concept erasure across random initializations while preserving non-target content and generation quality.
Under review at IEEE Conference on Secure and Trustworthy Machine Learning
A black-box embedding-aware adversarial prompting framework for attacking unlearned text-to-image diffusion models.
Sharif University of Technology
September 2023 – Present
GPA: 18.54 / 20
Supervisor: Dr. M. H. Rohban
Compositional Generation in T2I Diffusion Models
University of Guilan
September 2019 – August 2023
GPA: 19.58 / 20 · Ranked 1st
Supervisor: Dr. S. A. Mirroshandel
Medical Image Analysis
– Present
RIML Lab (Sharif University of Technology)
Supervisor: Dr. M. H. Rohban
Focus: Compositional Generation · T2I Diffusion Models · Visual Reasoning · Multimodal Models
–
University of Guilan
Supervisor: Dr. S. A. Mirroshandel
Focus: Medical Image Analysis
–
Instructor: Dr. Mohammad Amin Fazli
Led course coordination, quiz design, and grading; assisted with lectures, assignments, and student support.
–
Instructors: Dr. M. H. Rohban, Dr. M. Soleymani
Topics: Compositional Generation, Inference-time scaling, Test-time adaptation
Material preparation and grading.
–
Instructor: Dr. M. Soleymani
Topics: Generative Models, VAE, GAN, Diffusion
Quiz and exam design, and grading.
–
Instructor: Dr. M. H. Rohban
Topics: Diffusion and Generative Models
Supported exercises, exam/quiz design, and grading.
IDP
Dec 2025 · Overall 7.5
Listening 9 · Reading 7.0 · Writing 6.5 · Speaking 7.0
Direct admission to M.Sc. in Computer Engineering as an exceptional talented student.
–
University of Guilan · Highest GPA among all undergraduate Computer Engineering students · Certificate (Persian)
Tehran, Tehran Province, Iran