Federal Project Grant Award Summary

North Carolina State University received a $268,768 Project Grant from the National Science Foundation (NSF) Division of Information and Intelligent Systems under the Computer and Information Science and Engineering program (CFDA 47.070), awarded July 15, 2026, with completion targeted for June 30, 2029. This collaborative research initiative develops hierarchical optimization methods for processing ragged tensor operators used in deep learning workloads, particularly for artificial intelligence applications such as large language models and graph-based learning systems. The research addresses computational inefficiencies in current AI frameworks when processing irregular data with variable sizes and structures across tasks.

The deliverables include a unified framework for optimizing irregular tensor computations on graphics processing units through three integrated research thrusts: a multi-level auto-tuning system that dynamically selects execution strategies while minimizing runtime overhead; new communication and scheduling techniques that improve data locality and load balance across processing units; and compiler-level optimization strategies. Additionally, the project encompasses curriculum development in parallel computing and machine learning courses and community outreach initiatives designed to enhance workforce training in irregular data computation methods, supporting national priorities in healthcare, scientific discovery, and advanced manufacturing.