Irene’s research focuses on building scalable, energy-efficient infrastructure for large-scale machine learning training and serving workloads. As modern AI systems span thousands of accelerators and place increasing pressure on compute, memory, and network resources, she studies how to optimize distributed deep learning through the joint design of accelerator architecture, interconnects, and device placement. Rather than treating hardware and system components independently, her work develops holistic frameworks that reason across the full AI system stack to directly address the dominant performance and energy bottlenecks. Ultimately, her goal is to uncover new datacenter-scale designs that improve throughput per watt and enable more sustainable scaling of advanced AI models.
Irene Wang is a third-year PhD student at Georgia Tech, advised by Prof. Divya Mahajan. Her research focuses on systems for machine learning, with an emphasis on optimizing distributed deep learning infrastructure through co-design techniques. She has previously interned at Microsoft Research and Meta, and was recognized as an MLCommons Machine Learning and Systems Rising Star. Prior to joining Georgia Tech, she obtained her Bachelors of Applied Science from the University of British Columbia, where she was advised by Prof. Prashant Nair.