My research explores novel neuromorphic computing architectures using nanoscale memristors to overcome the memory-computation bottlenecks of CMOS-based AI hardware. Memristors, which retain resistance states and enable in-memory computing, offer high density, low power, and analog functionality ideal for matrix-vector multiplication. Building on recent advances in molecular memristors, this project will (1) design system-level architectures tailored to specific machine learning tasks, and (2) implement and characterize prototype systems to evaluate performance gains over conventional approaches, paving the way for energy-efficient, scalable AI hardware.

I am participating in the SuperUROP program to gain better research experience, deep dive into a next generation compute solution for our ever growing digital world, and become experienced in presenting and writing about research.