I am an undergraduate researcher at Duke University majoring in Computer Science and Statistical Science, with a focus on developing machine learning methods for protein modeling, molecular representation, and spatial transcriptomics. My research emphasizes architectural design choices in deep learning models that enhance how models extract information from biological sequences and molecular graphs.
In the Naderi Lab, I have contributed to several optimal transport (OT)-based deep learning methods, including a novel attention mechanism, a pooling method, and a graph neural network message-passing framework. This work prioritizes interpretable, geometry-aware representations and has been evaluated on biological and molecular benchmarks, including protein-related datasets.
Building on this foundation, I am currently developing an OT-driven architecture for zero-shot cross-platform cell alignment in spatial transcriptomics, utilizing gene expression and spatial coordinates. Together, these projects reflect my broader goal of advancing machine-learning-driven approaches to therapeutic discovery.
