A group of early-career engineers is tapping artificial intelligence to support Sandia’s digital engineering transformation. The three engineers, who have all been at Sandia for less than five years, put together a proposal and received funding from NNSA’s AI for Nuclear Security initiative for their project.

The overall goal of Sandia’s digital engineering transformation is to accelerate product delivery without sacrificing quality.

The AI project will improve workflows that benefit not just Sandia, but the entire nuclear security enterprise. It involves engineers from national security programs and nuclear deterrence programs.

“This is a great example of our early-career staff bringing fresh perspectives and a willingness to solve problems at a time when we are challenged to innovate and act with a sense of urgency,” said Brad Boswell, associate Labs director for Nuclear Deterrence Modernization and Stockpile Systems. “This mindset, coupled with the application of advanced tools puts us in a position to make critical impacts at a pace we simply haven’t been able to achieve previously.”

Model-based enterprise

The project, which aims to deliver four distinct tasks, will bridge the gap between model-based systems and definitions with AI-enabled tools.

“We envision tracking parts and systems all the way from requirements to production and testing,” said Catherine Appleby, an engineer who has worked at Sandia for about three years. “Having the full story that you’re able to reference in the future will be useful for surveillance and other activities.”

Catherine is referring to establishing the digital thread, a modern engineering approach that eliminates paper and integrates interconnected data and authoritative sources accessible by design and production agencies. It ensures both agencies work from the same information and data, often referred to as a single source of truth.

Model-based definitions

One goal of the project focuses on incorporating AI to reduce rework for current product definitions and support future model-based definitions, specifically by providing corrections to avoid costly delays.

“The designer will produce a product definition in a model-based definition,” said Max Anderton, an engineer in nuclear deterrence. “An AI assistant will review and correct mistakes, such as formatting, missing parameters and simple things that pass through the cracks during quality review, then slow down the lead time for these parts and assemblies.”

Max, who has been at Sandia for about three years, said his motivation for the proposal was creating efficiencies.

“I saw an opportunity for improvement, and I went for it,” he said.

Another task focuses on developing an AI tool that will quickly review and flag information in model-based definition files for sensitivities before submission to the classification office and release to the vendor. The AI review will not replace the normal classification review but will enhance and accelerate the process as more data is released to vendors.

“As we move into a more integrated digital thread environment, we want to define our parts by the model,” Max said. “The idea is that you can make it easier for manufacturers to make these parts by integrating more data into that model instead of just providing a 2D definition.”

Model-based systems engineering

On the model-based systems engineering side, systems engineer John Materum said he also saw opportunities for efficiencies.

“We have engineers who are reading a bunch of documents and thinking about how they can translate that into model-based systems engineering models,” said John, who has been a regular employee for about a year. “I felt that was a great opportunity to incorporate AI into their workflows.”

One goal focuses on having AI create elements in the models designed for the systems used in model-based systems engineering. The second goal is to analyze existing model-based systems engineering models in those systems.

“We want AI to query through that data, analyze it, validate it and give us feedback,” said John, who did coding work on model-based systems engineering programs while interning at Sandia.

At that time, Kahlil Stoltzfus was leading the AI tasks for model-based systems engineering. When he switched to a different job, he said it was clear John was the person to take the lead on these initiatives.

“Passion drives success, especially on these types of projects,” Kahlil said. “John had the technical expertise, and he had the passion to help drive it forward. It was a no-brainer that John would take over.”

Big impacts

Brad said with the collection of talent at Sandia, the time is ripe for transformation.

“We need Sandians to have the creativity to imagine the possible and have the courage to bring ideas forward — even small ones,” Brad said. “The convergence of new capabilities like AI and digital engineering have an opportunity to revolutionize the way we realize product.”

And this project has the potential to impact areas beyond nuclear deterrence.

“While it’s geared toward nuclear deterrence, it will help all projects across that use those tools,” Max said.

Sandia is working with Los Alamos National Laboratory, Oak Ridge National Laboratory and the Kansas City National Security Campus on these tasks over the next couple of years.

“People who are passionate will make this project successful,” Kahlil said. “If you have passion, you can learn the things you need to learn.”

And passion is something the early-career engineers leading this project have. You can hear it when they talk about their work.