MITRE's Federal AI Sandbox Unleashes AI's Potential for Public Good

MITRE operates six federally funded research and development centers (FFRDCs), helping various U.S. government agencies carry out their missions in innovative, efficient, and cost-effective ways. 

Because MITRE is a not-for-profit company prohibited from competing with industry, it often serves as a bridge between government and industry for the benefit of both. For example, government, industry, and academia trust MITRE to collect and analyze competitive data on MITRE platforms, such as the Federal AI Sandbox. This allows all parties to solve critical problems together as a community.

MITRE is collaborating with the National Oceanic and Atmospheric Administration (NOAA) and the National Weather Service (NWS) to develop new weather forecasting methods using the Federal AI Sandbox. Using NVIDIA Earth-2, an open platform for weather simulation, MITRE and NOAA are together advancing high-resolution spatial and temporal weather forecasts, which will improve public safety and national security.  

MITRE uses multiple NVIDIA technologies and platforms, including NVIDIA DGX SuperPOD and NVIDIA Omniverse, to evaluate, experiment with, and enhance weather forecasting models for oceanic regions. A research team uses and combines existing models to generate high-resolution forecasts and create visualizations in a custom application developed with NVIDIA Omniverse technologies. They leverage extensive satellite data and oceanographic measurements to develop 1-kilometer level precision, matching the accuracy of land-based predictions. This initiative promises significant advancements in maritime safety, shipping efficiency, and extreme weather preparedness, particularly for events like hurricane formation.

Further advancing ocean exploration, MITRE has developed a three-computer architecture by integrating AI on the edge, foundational model training on an NVIDIA DGX SuperPOD, NVIDIA Omniverse, simulation technologies. This setup enables seamless collaboration between autonomous systems, underwater vehicle testing in a highly instrumented tank, and large-scale AI model training in the Federal AI Sandbox powered by the DGX SuperPOD. Together, these systems create a virtuous loop of modeling, simulation, and real-world application to tackle complex challenges in areas like climate science and national security.

MITRE is also leveraging its DGX SuperPOD to address other challenges. For example, researchers are developing a groundbreaking foundational model to address the critical shortage of cybersecurity resources worldwide.

Another MITRE team is enhancing public benefits administration by creating sophisticated AI models to analyze data across multiple government departments. MITRE aims to optimize the processing of benefits claims, especially in areas where intricate regulations and diverse policies intersect. This forward-thinking approach has the potential to significantly reduce fraud and improve service delivery to citizens.

Philp added, “Our partnership with NVIDIA has been instrumental in advancing our AI initiatives. We learn from each other, leveraging NVIDIA's expertise to build on baseline models, collaborating on platform improvements, and expanding our existing field work to cover larger domains. This collaboration has accelerated our AI capabilities and opened new frontiers for simulation and innovation.”

“MITRE’s Federal AI Sandbox with the NVIDIA DGX SuperPOD is a powerful resource that we’re making available across the federal space; it ensures control over critical models and data while accelerating AI adoption across government agencies, driving innovations in national security, healthcare, and climate challenges,” said Philp.