Clearing a Path to Automating Insights from DoD Engineering Modeling and Simulation
Engineering modeling and simulation generate vast volumes of data—often fragmented across tools, studies, and teams. When aggregated at the project level, this data expands exponentially, making it difficult to navigate, reuse, or extract insight without the original study owner present. This presentation explores a structured approach to automating insight discovery from simulation data by leveraging metadata extraction and categorization techniques. By classifying data into context, inputs, and findings, we create a foundation for integrating large language models (LLMs) to support advanced querying and summarization. This enables new stakeholders—regardless of their technical background or role in the original study—to interpret and act on simulation results effectively. The approach addresses a known DoD pain point: reducing knowledge silos, accelerating decision-making, and supporting secure collaboration across DoD and with industry.  

This work transforms how DoD teams access and leverage simulation knowledge by automating the extraction and classification of metadata from complex modeling studies. By enabling large language models to understand and contextualize simulation data, the approach empowers broader teams—including those without direct involvement in the original analysis—to derive actionable insights. This significantly reduces dependency on individual experts, enhances cross-functional collaboration (within DoD and with contractors), and accelerates innovation cycles on engineering projects critical to national security.
IMPACT
This work transforms how DoD teams access and leverage simulation knowledge by automating the extraction and classification of metadata from complex modeling studies.
PRESENTER
Karasz, William
billy@rescale.com
516-613-1271
Rescale
CO-AUTHOR(S)
Crisup, Jeremy
jeremycrisup@rescale.com
Naas, Kiley
knaas@rescale.com
CATEGORY
Mod, Sim & Analysis for Decision Making
SYSTEM(S) USED
Rescale