Can Yaras

Pronounced Jahn YAH-rahs

cjyaras [at] umich [dot] edu

I'm interested in efficient, structured computation of deep learning primitives.

I will be joining Princeton Language and Intelligence as a Postdoctoral Research Fellow for Fall 2026. I completed by Ph.D. in Electrical and Computer Engineering at the University of Michigan, advised by Qing Qu and Laura Balzano. Before Michigan, I studied Electrical and Computer Engineering, Mathematics, and Computer Science at Duke.

Can Yaras standing in front of the Golden Gate Bridge — Taken by my good friend Sepehr Elahi

Select Publications

Full publication list

NeurIPS'25, Spotlight.

MonarchAttention: Zero-Shot Conversion to Fast, Hardware-Aware Structured Attention

C. Yaras, A.S. Xu, P. Abillama, C. Lee, L. Balzano

ICML'24, Oral.

Compressible Dynamics in Deep Overparameterized Low-Rank Learning & Adaptation

C. Yaras, P. Wang, L. Balzano, Q. Qu