Seminars

View all Seminars  |  Download ICal for this event

Building Efficient AI systems: A Cross-stack Approach to tackle AIs Compute, Memory, and Energy Demands

Series: CSA Faculty Colloquium

Speaker: Sanchari Sen, Assistant Professor, Department of Computer Science and Automation, IISc

Date/Time: Feb 06 16:00:00

Location: CSA Auditorium, (Room No. 104, Ground Floor)

Abstract:
Artificial Intelligence (AI) has become an integral part of our everyday lives. It Is powering everything from creative tools for art and text, to recommendation systems and chatbots. But behind these seemingly effortless interactions lies a massive amount of computing power and energy that often goes unnoticed.

As AI models have grown larger and smarter, they have also become extremely demanding on the computer systems that run them. In this talk, I will discuss a range of techniques across the hardware-software stack, aimed at addressing this challenge. I will specifically dive into the details of hardware-software specialization and sparsity-aware computing techniques. In the first half of the talk, I will present how specialization in IBMs Spyre accelerator and its accompanying software stack helps improve AI model execution. Following that, I will introduce SparCE, a set of lightweight micro-architectural and instruction set extensions that enable exploiting sparsity in general-purpose processor cores.

Speaker Bio:
Sanchari Sen is an Assistant Professor in the Department of Computer Science and Automation at IISc. Prior to this, she was a Senior Research Scientist at IBM T. J. Watson Research Center in New York. Her research focuses on hardware and software co-design techniques for efficient AI, including domain-specific accelerator and compiler designs. Sanchari holds a Ph.D. in Electrical and Computer Engineering from Purdue University, USA, and a B.Tech. from IIT Kharagpur, where she received the Institute Silver Medal. She has published over 25 papers in leading conferences and journals and holds several U.S. patents in the field of efficient AI computing. Her contributions have been recognized with multiple awards at IBM, including three Outstanding Research Division Achievement and two Outstanding Technical Achievement awards. At Purdue, she was awarded the Ross Fellowship and Bilsland Dissertation Fellowship for her research work.