USC+Amazon Center Announces 2024-2025 Projects and Fellows

The USC + Amazon Center is advancing its mission by announcing the three new students and projects it will be supporting for the 2024-2025 academic year.

Founded in January 2021 at the USC Viterbi School of Engineering, the USC + Amazon Center fosters collaborations between USC and Amazon researchers to develop new, privacy-preserving ML solutions. Since its inception, the center has supported over a dozen research projects examining the societal and industrial impacts of AI. It also provides funding to USC graduate students through its fellowship program, and offers its fellows the opportunity to be mentored by Amazon researchers.

The center is now one of the premier research centers in the new USC School of Advanced Computing, a unit of the USC Viterbi School and the university’s 23rd school.

“Our center is expanding its efforts to engage more faculty and students, reflecting the rapidly growing interest in trustworthy AI among researchers,” said Salman Avestimehr, inaugural director of the USC + Amazon Center. “This year, we are excited to support a diverse range of projects, from theoretical analysis to ML system design and practical applications, all aimed at advancing foundational research in trustworthy AI.”

The 2024-2025 fellows were selected through a rigorous nomination process that began earlier this year. Each will receive funding from the center and mentorship from Amazon scientists.

2024-2025 Amazon Fellows:

• Zeyu Liu, a second-year Ph.D. student in Electrical Engineering at USC, is advised by Professor Peter A. Beerel. His research focuses on developing efficient and responsible deep learning systems as part of the Energy Efficient Secure Sustainable Computing Group.
• Huihan Li, a third-year Ph.D. student in Computer Science at USC, is advised by Professor Xiang Ren. Her research on trustworthy natural language processing systems emphasizes improving generalization in language models and optimizing data selection for enhanced model performance. She holds a Master’s degree in Computer Science from Princeton University.
• Tejas Srinivasan, a fourth-year Ph.D. student in Computer Science at USC, is advised by Professor Jesse Thomason and works in the GLAMOR Lab. His research focuses on AI systems, and he has previously completed internships at the Allen Institute for AI and Microsoft Research. Before joining USC, he worked as a Research Scientist at the AI Foundation.

This upcoming year’s newly funded projects tackle pressing challenges in AI development, including sustainable model scaling and optimizing large language models (LLMs). 

Each year, through a competition process, the center provides support for several research projects focused on the development of new methodologies for secure and privacy-preserving machine learning solutions. All USC faculty are eligible to apply.

2024-2025 Projects:

• “Mechanistically-Aware Decoding for Ensuring Factuality During Retrieval Augmented Generation” by Professor Robin Jia, assistant professor in the Thomas Lord Department of Computer Science. Jia leads the AI, Language, Learning, Generalization, and Robustness (Allegro) Lab, focusing on building reliable natural language processing systems.
• “linearLLM: Enabling Fast LLM Serving via Attention on Top Principal Keys” by Professor Salman Avestimehr, Dean’s Professor in the Electrical & Computer Engineering and Computer Science Departments. He also serves as the director of the USC+Amazon Center.
• “Disaggregated Mixture of Experts Serving for Sustainable Model Scaling” by Professor Seo Jin Park, assistant professor in the Thomas Lord Department of Computer Science. Prior to joining USC, Park was a member of the Google Systems Research Group.

On Oct. 21, the USC + Amazon Center hosted its 4th Kickoff Meeting, featuring keynote speeches, presentations of the selected projects, and introductions on the new graduate student fellows. The event also gave guests the opportunity to speak with USC faculty and Amazon scientists. 

“This year’s projects build on the strengths of the Amazon-USC Center in responsible and trustworthy AI and advance important research questions from methods for reducing hallucinations to improving inference efficiency,” said Daniel Marcu, the vice president of Artificial General Intelligence at Amazon. 

“It was inspiring to be on the USC campus for the fall research kick-off event which brought together the newly awarded faculty and fellows with scientists from across Amazon’s Artificial General Intelligence team to launch these new collaborations,” Marcu added. 

Previous research funded by the center has included private labeling for voice assistants with cameras and decentralized learning. Over the past three years, the center has also supported several projects in deep learning, a powerful subset of machine learning for advanced AI applications such as image recognition, natural language processing and speech recognition. 

“It is great to see the excellent quality and wide range of research topics awarded in this year’s USC + Amazon competition,” said Yannis C. Yortsos, dean of the USC Viterbi School of Engineering. “I am confident that the research produced will drive the state of the art in these exciting fields.”

Published on December 5th, 2024

Last updated on December 5th, 2024