Electrical and Computer Engineering Seminar
Amir Yazdanbakhsh
Research Scientist, Google DeepMind
Wednesday, April 2, 2025, 10:00-11:00 am
B02 CSL Auditorium or Online via Zoom
Title: Co-design for Intelligence: Rethinking Abstractions for Scalable and Adaptive Systems
Abstract: The central theme of this talk is the importance of co-design—tightly integrating hardware, software, and machine learning, while rethinking long-standing system abstractions—to build the next generation of scalable and adaptive intelligent systems. Along this journey, I will share several techniques I’ve developed throughout my career to improve the efficiency of serving generative AI models and warehouse-scale computing. I’ll take a deeper dive into a recent work that demonstrates how to extract performance gains in today’s post-Moore era. In particular, I’ll highlight the rising importance of code optimization, the limitations of traditional approaches, and how Generative AI opens up new opportunities for complex, high-level code transformations. I’ll conclude by outlining open questions and promising directions that I believe will shape the future of intelligent systems.
Amir Yazdanbakhsh (https://www.ayazdan.com/) is a Research Scientist at Google DeepMind, where he works at the intersection of machine learning and computer architecture. Before joining DeepMind, he completed the Google Brain AI Residency, where he applied machine learning to accelerator design and developed scalable reinforcement learning systems for complex real-world tasks. His work has driven key innovations in hardware-software co-design, leading to multiple headline hardware features across several generations of Google TPUs. Dr. Yazdanbakhsh is an ISCA Hall of Fame inductee, and his research has earned an IEEE Micro Top Picks distinction—one of the first machine learning papers to receive this honor—along with two honorable mentions, a CACM Research Highlights nominations, a Distinguished Paper Award at HPCA 2016, and a Distinguished Artifact Award at ISCA 2024. He received his PhD in Computer Science from Georgia Tech and was awarded the Microsoft Research PhD Fellowship and the Qualcomm Innovation Fellowship.