Research

Where this is headed

My research interests are centered around how AI systems learn useful representations of the world. With a background in computer vision, I’m particularly interested in world models, representation learning, and multimodal systems, and in how we can move beyond treating different forms of information as simply collections of tokens. I’m interested in understanding how models can build more structured and efficient representations, learn from different modalities, and develop a richer understanding of the environments they interact with.

Research work

AI Researcher — University of Isfahan · Sep 2022 – Jul 2023

Technical report (PDF, 19 pages) ↗

Comparative deep learning analysis — ViT vs. CNNs for outdoor fire detection

Engineered a computer vision pipeline benchmarking attention-based architectures against convolutional baselines for early fire and smoke detection in unstructured outdoor environments, aimed at reducing disaster-response latency. I designed the benchmark, curated the dataset, and ran the five-architecture comparison.

Outcome. Established the transformer as the stronger detector at 95.7% accuracy with 13 false positives, against 30–46 for the CNN baselines, while identifying RegNetX as the better fit for aerial smoke. Diagnostic analysis on every false positive traced 35% of errors to firefighter and red-truck features in the images.

Read the full write-up →

PythonPyTorchCNNsVision Transformerstransfer learningcomputer visionNumPymodel benchmarking

AI Engineer — Behyar Sanaat Sepahan · Jun 2021 – Jun 2022

Deep-learning edge-preserving CT reconstruction

Engineered a deep learning pipeline to reconstruct diagnostic-quality medical images from low-dose CT scans, easing the trade-off between patient radiation exposure and diagnostic clarity. I owned the model architecture, loss design, and the synthetic training-data pipeline.

Outcome. Delivered a reconstruction model that preserved diagnostic edge detail where pixel-loss-only training blurred it — trained without access to paired clinical scans, using a hybrid perceptual/pixel loss and a differentiable Sobel edge layer.

Read the full write-up →

PythonTensorFlow/KerasCNNsautoencoderstransfer learningcomputer visionNumPymedical imaging