Two University of California, Merced computer science and engineering professors take a deep dive into artificial intelligence in a project with Amazon.
Dong Li and Xiaoyi Lu were the recipients of the tech giant’s recently announced Amazon Research Award.
Amazon Research Awards provide unlimited funding and Amazon Web Services promotional credits to academic researchers studying research topics in multiple fields. Li and Lu are among 63 recipients representing 41 universities from eight countries in this grant cycle.
Both projects work in conjunction with AWS Trainium, a chip dedicated to high-performance deep learning training of generative AI models.
Li’s winning proposal, “Efficient Sparse Training with Adaptive Expert Parallelism on AWS Trainium,” aims to find ways for different computers to learn faster and consume less power.
Lu’s project, “Accelerating Large-Scale Language and Inference Model Workloads with AWS Trainium,” aims to speed up computer processes. He will be researching how to train cutting-edge AI models, such as those used in OpenAI GPT and Google Gemini, with Trainium. His project studies memory systems and communication efficiency to enable the training of high-performance AI models.
Amazon said the proposals were reviewed for the quality of their scientific content and their potential to impact the research community and society. Additionally, Amazon encourages the publication of research results, the presentation of research at Amazon offices around the world, and the release of related code under open source licenses.
Recipients have access to over 700 Amazon public datasets and can use AWS AI/ML services and tools through AWS Promotional Credits. Each recipient will be assigned an Amazon researcher to provide consultation and advice, as well as the opportunity to attend Amazon events and training sessions.
“Academic AI researchers face a fundamental challenge: To advance machine learning research and educate the next generation, they need access to powerful, affordable, cutting-edge infrastructure,” said Yida Wang, Principal Applied Scientist at AWS AI.
“The Build on Trainium program directly addresses this barrier.…Build on Trainium represents AWS’s commitment to democratizing AI research through collaborative partnerships with academia. We foster an environment where researchers experiment freely, students learn on production-scale infrastructure, and academic innovation shapes the future of machine learning for everyone.”
Both researchers said they were excited about the possibilities the grant offers.
“By accelerating large-scale language and inference model workloads on AWS Trainium, this project aims to push the performance boundaries of modern AI systems and make advanced AI capabilities more efficient and accessible,” said Lu.
Lee said the award recognizes the quality of high-performance computing research at UC Merced.
“This award will give our students access to state-of-the-art hardware on AWS for their AI research, allowing us to offer them interesting research projects, internships, and more,” he said. “We will also connect AWS and UC Merced for technology, education, and business collaboration.”
