Training Tomorrow’s Data Scientists

When Melanie Segado wanted to investigate the brain activity of cello players for her neuroscience PhD, she encountered a problem: Magnetic resonance imaging (MRI) was the best available tool to visualize neural activation, but MRI machines require participants to lay motionless in a confined space. To make her study work, she helped develop an experimental set-up that allowed cellists to play their instruments within the confines of an MRI machine.

As Segado watched master cellists quickly adapt to the strange set-up, it sparked a new interest in how humans learn to move—an interest that brought her to Penn in 2023, where she now works as a postdoctoral researcher in the lab of neuroscientist Konrad Kording. She’s one of the postdoctoral researchers supported by the Data Driven Discovery Initiative (DDDI), which began offering this type of fellowship in spring 2025 as one facet of Penn Arts & Sciences’ broader initiative to pioneer advancements in AI research, applications, and education.

The program is rapidly growing. In its first years it had only eight fellows. This year, as part of its new AI x Science initiative, DDDI has expanded the program to include fellows from Penn’s School of Engineering and Applied Science, led by Rene Vidal, and the Perelman School of Medicine (PSOM), led by Marylyn Ritchie and Li Shen. And at the beginning of May, three new postdoctoral researchers from Penn Arts & Sciences also joined DDDI, bringing the total number to 25.

Bhuvnesh Jain, Co-Director of DDDI and Walter H. and Leonore C. Annenberg Professor in the Natural Sciences, says that having a robust postdoctoral research program is integral to DDDI’s mission of sparking discoveries through data science at Penn. Jain co-founded the DDDI postdoctoral program with Greg Ridgeway, Rebecca W. Bushnell Professor of Criminology and a professor of statistics and data science. “Particularly in the application of AI for scientific inquiry, postdocs are really leading the charge,” Jain says. “They are current on the rapidly evolving methodology, and have the confidence to explore new ideas and collaborations.”

All postdoctoral fellows have a home laboratory in their respective schools, but receive additional monetary support from DDDI and meet with others in the program for weekly data science lunches, where they learn about new techniques and research from professors across the University. Other events include an AI seminar series, as well as an annual summer data science hangout designed to connect postdoctoral researchers with undergraduate students in their field.

Colin Twomey, executive director of DDDI, was one of the first DDDI postdoctoral researchers when the program launched in the fall of 2021; for his research, he integrated data across linguistics, biology, and psychology to explore how different cultures use language to define colors. He says an opportunity like the fellowship offers individuals a place to discuss new research methodology, troubleshoot problems, and explore the work of their peers.

“It’s really an exciting time to have these kinds of interdisciplinary exchanges. There’s a lot to be discovered in so many different fields,” he says, adding, “Having a community of postdocs and faculty to draw inspiration from is really meaningful.”

New AI x Science Fellows

Segado is one of 11 inaugural Penn AI x Science Postdoctoral Fellows, a subset of the DDDI postdoctoral program that fosters research at the intersection of science, machine learning, and artificial intelligence across Penn Arts & Sciences and Penn Engineering, in partnership with Engineering’s Innovation in Data Engineering and Science (IDEAS).

AI x Science was born out of a desire to expand DDDI’s postdoctoral program amidst the evolving landscape of research and artificial intelligence. As AI innovation increasingly shifts from academia to industry, the fellowship was a way for Penn to maintain a leadership role by focusing on fundamental questions and long-term impact, according to Jain. Penn’s tightly integrated campus and strong interdepartmental relationships make it uniquely suited to support this kind of program, he adds, and it ultimately aims to reshape the postdoctoral experience by offering structured peer engagement, dual mentorship, and opportunities for interdisciplinary exploration—going beyond the typical, siloed postdoc trajectory.

Image

Segado has already experienced the benefit of such a structure. In her research, she is using AI to develop algorithms that can screen for cerebral palsy in infants based on visual data. Her models track infants in videos and analyze their movements to determine whether their motor development appears abnormal. The research could help support clinicians and promote earlier diagnoses for movement disorders. She says that collaborating with data scientists across different fields has helped her learn new research methods and applications for her work.

This spring, DDDI also introduced the AI x Science x Medicine fellowship, which brings together five new postdoctoral researchers from PSOM applying AI tools to solve today’s biggest health issues. That includes Tess Cherlin, who works in the lab of Shefali Setia-Verma, a Penn Medicine Assistant Professor in the Department of Pathology and Laboratory Medicine and at the Institute of Biomedical Informatics. Cherlin is using machine learning methods to develop better diagnostics for women’s health issues such as polycystic ovary syndrome, a disease that affects an estimated 116 million people worldwide but remains undiagnosed in up to 70 percent of cases.

“It’s a very logical fit for AI,” Cherlin says. “We want to show that using AI is just as accurate, if not more so, than what already exists in the field.”

An Interdisciplinary Community

Cherlin joined Penn in 2022 as part of a teaching-focused fellowship, and her AI x Science x Medicine fellowship will give her a chance to continue honing that skill. This summer, she’ll lead a workshop training undergraduates about data science methods in health research as part of DDDI’s annual summer data science hangouts.

She says she’s excited to see how the students envision using this in their own fields. “I’ll be sharing a technique that’s really useful in the medical health field, but these students are from all over the University,” Cherlin says. “I’m really curious to see what they all do.”

As a postdoc, you’re often coming in sort of isolated. Having a community of postdocs has been fantastic. I’ve never been in a space before where it’s so easy to talk to someone from physics or astronomy—that is really rare, and it’s kind of the ideal academic environment.

That interdisciplinary spirit is core to DDDI’s mission, and it’s part of what makes it such a cohesive community, says Vlad Ayzenberg, who will join Temple University as an assistant professor this coming fall and was a former fellow in the lab of Michael Arcaro, an Assistant Professor of Psychology. Ayzenberg says being part of DDDI helped him feel connected to the broader Penn community.

“As a postdoc, you’re often coming in sort of isolated,” he says. “Having a community of postdocs has been fantastic. I’ve never been in a space before where it’s so easy to talk to someone from physics or astronomy—that is really rare, and it’s kind of the ideal academic environment.”

Segado adds that she feels a similar sense of collaboration when she talks about her research with her DDDI peers. 

“I’m adapting pre-trained models to work with small, specialized datasets,” she says. “That’s something that people in astrophysics, genomics, and other domains of medicine are also working on. Putting all of us in a room together where we can share our research trajectories, our triumphs, and our struggles has been really helpful for thinking outside the box.”

For Jain, a veteran astrophysics and cosmology researcher, it’s also rewarding to lead a program that brings together scientists from so many far-reaching disciplines. “For much of my career, I was part of the century-long trend in academia of greater and narrower specialization,” he says. “It’s really amazing to see some opening up and crossing of disciplines that hasn’t happened in many decades.”