Though Scott Stewart recently received an Early Career Award from the Institute of Nuclear Material Management, he is regarded as a seasoned professional in the nuclear field with over 10 years of experience. Recognizing the importance of bringing new ideas to the field, Stewart seeks to attract the next generation of nuclear scientists and engineers by making connections between data science and nonproliferation.
“We need to reach out, train the next generation and build supportive networks,” said Stewart, a nuclear safeguards engineer at ORNL. “This is the type of profession where 10 to 15 years of experience to understand the context can make an impact.”
In his award acceptance speech, Stewart highlighted the small nonproliferation community in the United States. People typically are either introduced to this field, or they stumble across it, he said. “But to advance this national security priority, more young people need to be exposed to the opportunities within this field.”
After completing an undergraduate degree in physics with a minor in public service, he thought the nuclear Navy was his calling. Instead, he was introduced to neutron measurement techniques used in nonproliferation research thanks to the mentorship he received as a research assistant at Los Alamos National Laboratory. As he explored this niche, the amalgamation of policy and political science with technical expertise “hooked me in,” Stewart said.
The dance between technical capability and policy implementation now drives Stewart to find cutting-edge methods for helping nuclear organizations achieve their mission. For instance, what does it mean to have a high degree of confidence that nuclear material hasn’t been diverted from peaceful purposes? To answer this question, he wants to increase confidence through better data analysis while providing useful updates to the tools inspectors use in the field to gather information.
“Nonproliferation-relevant data is sparse,” Stewart said. “We’re looking for disparate, hard-to-detect events. When you’re making conclusions with any sort of data-driven approach, like a machine learning algorithm, you need to have high confidence in what it’s doing and why it’s doing it.”
Stewart has been passionate about data science for nonproliferation over the past five years. He builds algorithms to look at nonproliferation-relevant data in novel ways to get the most out of often-small datasets and develops metrics that consider nonproliferation policy objectives when evaluating algorithms. One of Stewart’s approaches, called the decision framework, used mathematical constructs to represent policy objectives and integrate considerations such as cost and effectiveness. While it might be common to evaluate an algorithm based on metrics such as accuracy, there is no guarantee that the most “accurate” algorithm is the best fit for a specific application. By considering multiple factors, including policy requirements, as a basis for evaluation, the decision framework helps more effectively and objectively compare data science approaches to select the most appropriate approach or algorithm to meet a specific policy objective.
“We created a methodology for evaluating an algorithm against a set of end user objectives to determine whether the algorithm will meet those objectives,” Stewart said. “In addition to helping bake those objectives in, the method allows for apples-to-apples comparisons when looking at different algorithmic approaches to a problem.”
While advancing his own science, Stewart continues to focus on attracting data scientists to the nonproliferation field. He co-founded a data science working group within INMM and organized a hackathon in 2022 to introduce students to data science.
As he looks to the future, Stewart hopes to bring a higher degree of efficiency to international inspectors through data analytics and natural language processing. Along the way, he intends to develop a cadre of peers ready for long, fulfilling careers promoting and ensuring the peaceful use of nuclear material.
“To help the International Atomic Energy Agency remain effective and efficient, we need people to build decades of experience with depth in technical solutions,” Stewart said.
UT-Battelle manages ORNL for the Department of Energy’s Office of Science, the single largest supporter of basic research in the physical sciences in the United States. The Office of Science is working to address some of the most pressing challenges of our time. For more information, please visit energy.gov/science. — Liz Neunsinger