I am a fourth year PhD student in Computer Science at UIUC advised by Prof. Chris Fletcher. I work at the interesection of computer architecture and machine learning: I have designed high-performance processors targeted at different ML workloads, I have built ML tools to make computers more efficient.
Before joining CS@Illinois, I worked at ARM, where I had a fantastic opportunity to build real-world products as a design engineer and develop successful prototypes to drive new products as a researcher in industrial R&D. I received my bachelors from NIT-K Surathkal in 2015 from the ECE department, where I was advised by Prof. Sumam David.
I am always looking for collaborations, especially motivated undergrads/grads. If you are interested in collaborating, or like any of my prior works and have ideas to advance it, or if you simply want to have a chat, feel free to contact me!
Ph.D. in Computer Science, 2017-Present
University of Illinois at Urbana-Champaign
B.Tech in Electronics and Communication Engineering, 2011-2015
National Institute of Technology Karnataka, Surathkal
\n\n Mind Mappings is a first-of-its-kind tool for mapping space search for hardware accelerators using gradient descent. \n\n
ExTensor is an accelerator for Sparse Tensor Algebra
Hardware and Software Infrastructure for Machine Learning group
Mind Mappings is a novel framework that enables first-order optimization with gradient descent for mapping space search, a core challenge in deploying efficient programmable accelerators.
ExTensor is an acceletor for Sparse Tensor Algebra, a key class of workloads that powers crucial areas such as deep learning. Key insight behind the design is to hierarchically eliminate ineffectual work that exists due to sparsity to demonstrate significant speed-ups.
Morph is a flexible hardware accelerator for 3D-CNNs, a key workload used in video understanding. Key insight behind the design is to design a flexible hardware that allows high degrees of flexibility that allows different mappings of 3D-CNNs that maximizes performance for the given layer of 3D-CNN..
A result of reducing precision in deep neural networks is increasing repetition of parameters in DNN models. UCNN is a hardware-software codesign approach based on algebraic reassociation techniques to exploit weight repetition that significantly improves efficiency of DNN inference.