Federal Project Grant Award Summary

George Mason University received a $362,095 Project Grant awarded July 15, 2026, through the National Science Foundation's Division of Information and Intelligent Systems under the Computer and Information Science and Engineering program (CFDA 47.070). The collaborative research project, titled "Hierarchical Optimization of Ragged Tensor Operators for Deep Learning Workloads," addresses computational inefficiencies in artificial intelligence (AI) applications that process irregular data structures, such as those used in large language models and graph-based learning systems. The three-year project (completion June 30, 2029) will develop a unified framework for optimizing irregular tensor computations on graphics processing units (GPUs), featuring a multi-level auto-tuning system for dynamic execution strategy selection, enhanced communication and scheduling techniques to improve data locality and load balancing, and compiler-level optimizations for scalable execution across single-device and multi-device platforms.

Beyond core research deliverables, the project includes curriculum development in parallel computing and machine learning courses and community outreach initiatives to enhance workforce training in irregular data computation methods. These educational and training outcomes align with NSF's emphasis on advancing both research frontiers and human capital development in computing disciplines. The research directly supports national priorities in healthcare, scientific discovery, and advanced manufacturing by enabling faster, more cost-effective, and energy-efficient AI systems capable of handling real-world irregular workloads that current AI frameworks struggle to process efficiently.