Anyone who’s spent an unintended hour scrolling on their phone knows how powerful algorithms can be; those mathematical instructions guide computers and other tech in their autonomous decision making—from what Instagram photo to show next to the optimal temperature of a pharmaceutical reactor.
But, just like people, algorithms sometimes have difficulty making good choices, especially when faced with incomplete data or uncertainty. Joel Paulson, who joined the Department of Chemical and Biological Engineering in August 2025 as the Gerald and Louise Battist Associate Professor, uses AI and machine learning techniques to make algorithms that improve predictive models and process optimization.
The work has wide-ranging applications; so far he’s applied his techniques to optimize pharmaceutical and semiconductor manufacturing, smart building management, plasma jets, colloidal self-assembly, alloy design, and drug design.
In many ways, Paulson is the ultimate team player, using his techniques to improve processes for other researchers. “I’m very application-forward, but I’m pretty agnostic; we work with all kinds of people,” he says. “I’m not the expert in their domain, but I know how to basically accelerate and improve things in their domain, automate more, and get stuff off their plate. That’s how I look at it.”
After studying chemical engineering as an undergraduate at the University of Texas at Austin, Paulson earned his PhD at MIT, where he focused primarily on stochastic process control for pharmaceuticals, developing systems that could account for uncertainty. Later, as a postdoctoral scholar at the University of California, Berkeley, he worked on process control for semiconductor etching in collaboration with the company LAM Research. There, he became more interested in machine learning and its potential role in process control.
As a faculty member in the Department of Chemical and Biomolecular Engineering at The Ohio State University, Paulson has pursued AI and machine-learning informed process control for engineering and chemistry-related research. That work has earned him many honors, including a National Science Foundation CAREER award, the AiChE 35 under 35 Award, the OSU Lumley Research Award and the David C. McCarthy Engineering Teaching Award.
Most recently, Paulson has turned his work to molecular design, finding ways to speed up and optimize the discovery of alloys and materials on extremely tiny scales. Some of his most successful work, he says, has been collaborating with chemists seeking to design new sustainable battery materials.
At UW-Madison, he wants take things up a notch. “I want to take those ideas and start doing more high-throughput research. I want to start doing automated design, self-driving systems for molecular design and materials more broadly,” he says. “Systems where a human doesn’t always have to be fully in the loop. That’s what I’m really excited about as I start at UW-Madison.”
He’s hoping to move into areas that include developing materials for batteries, optimizing catalysis to find new energy-production chemistries, and designing drugs and proteins for potential cancer therapies. But he says he’s open to working with the huge variety of projects and collaborators he’s likely to encounter at the university.
Over time, Paulson hopes he can take the various and wide-ranging optimization methods he’s developed and synthesize them into a single unified AI system, a sort of ChatGPT for process control and optimization. “People could formulate ideas using natural language, like in large language models, then use our algorithms at the back end to connect them together and achieve a goal,” he says.
Paulson says he’s also excited to share his insights with students. While some much of his work can seem abstract, he says he tries to make things as concrete as he can for students. That means using hands-on projects to teach optimization, like building an energy management system then applying process control and optimization principles in context.
“I like the challenge of trying to present these ideas to people,” he says. “I love optimization and efficiency, so I’m always trying to figure out the best way to compress this stuff so people learn as effectively as possible.”