Backstory
Drawn ever deeper into the veiled Mysteries of Vision Transformers—arcane engines binding sight and mind—Amir seeks to unearth forgotten patterns from the buried magics of tensor and manifold. In parallel with his solitary studies, he has pledged service to the Guild of Neuromatch, mapping the labyrinths of the living brain, and to the House of Digipay, where he forges automata to test and temper the merchant magics of the realm. A traveler between the realms of mind, machine, and myth, Amir weaves his sigils across the shifting firmament of knowledge, etching new runes upon the dark and towering walls of human understanding.
Master's Degree
Computer Science
Tarbiat Modares University (TMU)
2022 – Present
GPA 4.0/4.0 | Rank 1st
Bachelor's Degree
Computer Science
Amirkabir University of Technology (AUT)
2017 – 2021
Research Journey
Impact Scholar of Neuromatch
Ventured alongside a wonderful fellowship into the labyrinths of brain and machine, studying nonlinear CEBRA and ancient spatial filters to uncover hidden patterns of the mind.
Architect of Intelligent Tests
Led a pioneering guild at the House of Digipay, where I and my trusted team forged enchanted frameworks using LLMs and Visioncraft, crafting automated rites to test and safeguard merchant magics.
Contributing Sage at DeepTenLab
As a long-standing contributor to the DeepTenLab guild, I shaped modular spellbooks for tensor and manifold magic, guiding the weaving of visions, scrolls, and brainwaves alike.
Teaching Missions
Assistant at Neuromatch Academy
Assisted scholars on their quest through Computational Neuroscience, supporting their learning and project crafting under the great halls of Neuromatch.
Assistant to the Deep Learning Scribes
Aided seekers in deciphering the tomes of Deep Learning at the Citadel of Tarbiat Modares.
Assistant to the Machine Learning Circle
Offered guidance and crafted challenges for apprentices studying Machine Learning within TMU's hidden chambers.
Assistant to the Order of Special Topics
Supported learning rites across rare and evolving domains of Data Science.
Assistant in the First Scrolls of Data Mining
Guided fledgling miners through the deep veins of data within Amirkabir Keep's ancient halls.
Other Ventures
Software Test Enchanter
At the House of Digipay, I wove a tapestry of manual trials and automated incantations, ensuring the resilience of merchant systems through test enchantments guarding payment spells, user rites, and merchant contracts.
Hierarchical Local-Global Tensor Layer (HLGTL) in Deep Convolutional Neural Networks
2025 [Under Review]
Explores the design of a new tensor-based layer, blending local and global features, to enhance deep convolutional architectures.
Submitted to: Expert Systems With Applications
A Comparative Analysis of Spatial Filtering and CEBRA Methodologies for Enhancing BCI Adaptability on ECoG Data
2025 [Micropublication]
This study compares Linear Spatial Filtering and CEBRA for processing ECoG data in BCIs, highlighting that while spatial filtering improves SNR and CEBRA captures latent patterns, each method falls short in classification performance or generalizability. The results underscore the need for hybrid or enhanced processing techniques to improve BCI adaptability across individuals.
Submitted to Zenodo: Sciencematch Community
Negative Scoring Within Attention Mechanisms
Investigating the introduction of negative scoring schemes within transformer attention to enhance representational focus.
Local-Global Hybrid Vision Transformers for Small-Scale Datasets
Designing vision architectures that blend local and global feature modeling, optimized for limited data realms.
Multi-Level Laplacian Eigenmaps
Developing hierarchical spectral methods for advanced graph and manifold learning tasks.