Mathilde Papillon, a Ph.D. candidate in the Geometric Intelligence Lab at UC Santa Barbara, serves as an AI Research Scientist for the Los Angeles Dodgers.
When the Los Angeles Dodgers stormed the field to celebrate their 2025 World Series victory, the vibe was, in the words of Mathilde Papillon, “off the chains.”
But Papillon, a PhD candidate in the Department of Physics at UC Santa Barbara, views the game through a lens that captures much more than just the final score. Where fans see a curveball or a swing, Papillon sees geometry. She sees vectors, manifolds and high-dimensional data points.
“I spend much more time looking at stick figure-like reconstructions of baseball motion on my computer than I do of live human bodies doing baseball,” Papillon said. “I’m always imagining the stick figure overlaid on top.”
Papillon occupies a unique position in the history of the franchise. Hired in the spring of 2025, she is the Dodgers’ first AI Research Scientist. Working within the Department of Baseball Innovation, a small, specialized unit within the front office, she applies the complex mathematics of her doctoral research to the physical realities of the diamond.
Her work represents what she calls “Moneyball 2.0” — a shift from analyzing static statistics like batting averages to decoding the massive, complex datasets generated by modern motion capture technology.
From choreography to code
Papillon’s path to Major League Baseball was anything but linear. Before she was a physicist, she was a dancer.
“I trained as a professional dancer before coming to graduate school,” she said.
When she arrived at UCSB, originally intending to work in condensed matter physics, she felt a pull to return to her roots in human movement. That instinct led her to the Geometric Intelligence Lab, run by Nina Miolane, an assistant professor in the Department of Electrical and Computer Engineering.
In their first meeting, Papillon made an unconventional proposal.
“I still can’t believe I said this, but I was like, ‘Can we build something for dance?’” Papillon recalled. “And she said, ‘What?’ And I said, ‘I want to do dance.’”
Miolane’s response was immediate.
“She just got up on the whiteboard and she started drawing these little skeletons as input data, output data,” Papillon said.
That meeting set the trajectory for her research: using AI to understand how the human body moves. Papillon explains her work by comparing it to large language models.
“The way that I explain my work is like ChatGPT, but instead of sending in words and sentences and getting back words and sentences, I send in movement and I get back movement,” she said.
Papillon maps out the neural network architectures she builds in the Geometric Intelligence Lab to model complex human motion. Her research treats the human body not just as isolated joints, but as a connected, functional system.
The shape of the pitch
“I build neural networks that are more similar to how we as humans perceive motion, compared to the traditional computer vision approaches,” Papillon said.
This nuance is critical in baseball. Modern stadiums are equipped with systems like Hawk-Eye, which track the position of players’ joints at every frame of the game. This generates a “mountain of data” that teams are racing to understand.
By applying neural networks for human motion, Papillon helps the Dodgers look for patterns the human eye might miss. The goal isn’t to replace the intuition of coaches or scouts, but to give them a new tool.
“We’re not hoping that AI is going to think for us and replace us, but serve as an extra tool that can give us new ideas and new perspectives,” she explained.
In the Geometric Intelligence Lab, Papillon diagrams the intersection of physics and artificial intelligence. Her work applies physics-based computations — like velocity and momentum — to help AI models understand the mechanics of dance, pedestrian motion, basketball and general human motion.
A lab of one’s own
The transition from an academic lab at UCSB to a Major League front office was seamless, largely due to the research culture she experienced on campus.
“Experiencing research in general, tackling tough problems, failing at times, equipped me with a whole lot of resilience and ingenuity,” Papillon said.
Miolane, Papillon’s graduate advisor, attributes this success to a rare combination of skills.
“Mathilde was able to connect her intuitive sense of movement, that she had from dance, to the geometric models we use in AI,” Miolane said. “Very few people are equipped to turn such insight into new mathematical models for AI — but she did. It’s been incredible to witness her creating a whole new paradigm for AI in sports.”
Papillon credits the interdisciplinary nature of UCSB — where a physics student can work in an engineering lab on dance choreography — for preparing her to innovate in sports.
“I am really grateful to have had, and to continue to have the ideal playground for learning and building stuff at UCSB. And that translates right over to learning and building things for sports teams,” she said.
The future of the game
While fans worry that analytics might make baseball boring, Papillon argues the opposite. As the game evolves — with pitch clocks and smaller bases — the modeling must evolve with it. “It's not a static thing. It’s not like an AI model trying to beat Go or beat chess. The idea is not to build a supercomputer that will dominate the game forever,” she said.
Instead, her work is focused on perfecting the understanding of the game and, crucially, keeping players’ performance as optimal as it can be.
As she works towards her PhD, Papillon continues to split her time between her academic research and her role with the Dodgers. It is a demanding schedule, but one that places her at the forefront of a new era in sports science.
And while she may not have been a die-hard baseball fan when she started, the World Series run left an impression.
“It was really cool to get to witness the high stakes of the game, especially having gotten a front-row seat to all the quantitative work that goes on behind the scenes,” she said.