The National Science Foundation Division of Computing and Communication Foundations awarded Northeastern University $599,921 on January 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop hardware-algorithm co-design approaches that enhance the reliability, efficiency, and sustainability of deep neural networks deployed in safety-critical and life-critical applications.
The project addresses failures and design bugs observed during AI system deployment that compromise hardware security, system sustainability, and operational trustworthiness. Northeastern will create end-to-end approaches to understand the impact of hardware failures and bugs on advanced deep learning workloads, develop mitigation strategies through co-design, and incorporate user studies to explore adaptive architectural solutions. The work draws on computer architecture, robust system design, and machine learning disciplines. The recipient will also enhance computer architecture education through visualization and workshops targeting high school students and facilitate technology transfer.
Performance runs through December 31, 2028, at Northeastern University in Boston, Massachusetts. The award is structured as a Project Grant under the CAREER program framework.