AMSC Ph.D. student Shashank Sule (M.S. ’24, applied mathematics & statistics, and scientific computation) helps develop methods that could improve brain disease diagnostics and molecular simulations.
Shashank Sule (M.S. ’24, applied mathematics & statistics, and scientific computation) considers himself a latecomer to math. As an Amherst College freshman in 2016, he wanted to pursue chemistry research but wasn’t accepted into his preferred lab.
Undeterred, he enrolled in a Fourier analysis math class to learn the basic tenets of Fourier transform infrared spectroscopy, an analysis technique used in chemistry, to bolster his skill set before reapplying. He didn’t expect this math class to change his plans completely. “I took that class and thought, ‘Okay, I don't want to major in chemistry anymore. I'd like to do mathematics instead,’” Sule recalled. “I liked how a lot of the arguments in Fourier analysis fit together elegantly and how far-reaching the applications of that subject were.”
Sule has been studying math ever since and is now a Ph.D. student in the University of Maryland’s applied mathematics & statistics, and scientific computation (AMSC) program. With applications in medical imaging and drug discovery, Sule’s research blends machine learning with mathematical techniques like applied harmonic analysis—a method that includes Fourier analysis and breaks functions down into smaller pieces for further study. Sule’s love of dissecting problems has aided his research, which aims to identify techniques for improving brain disease diagnostics and molecular simulations.
“I've always had a tendency to deconstruct or analyze things, and I also liked seeing how arguments fit together,” Sule said. “I think that is something that attracted me to math in the first place.”
Though Sule pursued math later than some of his peers, a study abroad program in fall 2018 helped to bring him up to speed. As an undergraduate student, he spent a semester learning from Hungarian mathematicians in Budapest—a humbling and rewarding experience, according to Sule. “They’re famous for their problem-solving culture, so there were a lot of Olympiad-style problems,” Sule said. “I wasn’t used to that since I wasn't in the math world as a kid, but at some point, I started getting the hang of it and it became really fun.”
When Sule returned to Amherst, his advisor, University of Maryland alum Karamatou Yacoubou Djima (M.S. ’11, Ph.D. ’15, applied mathematics & statistics, and scientific computation) suggested that he focus on harmonic analysis for his undergraduate thesis. Sule was instantly captivated by the field’s applications in everything from medical imaging to machine learning—so much so that he decided to keep pursuing this research after earning his bachelor’s degree in mathematics in 2020. He chose UMD’s AMSC program because of his advisor’s recommendation and the Norbert Wiener Center for Harmonic Analysis. Since joining UMD, Sule has had the chance to explore both sides of his mathematical interests—pure and applied. “What gives me a lot of joy is finding good proofs but also finding mathematical arguments that lend themselves well to computation,” Sule said. “I think that's a really interesting side of harmonic analysis.”
In his current research, Sule works closely with his co-advisors, Mathematics Professors Wojciech Czaja and Maria Cameron. With Czaja, he uses a mix of classical and deep learning methods to develop better methods of identifying changes in myelin—a protective layer around nerve fibers and an important biomarker for brain disease. “If you want to diagnose Alzheimer's, traumatic brain injury or multiple sclerosis, there is this component in the brain called myelin, and researchers want to measure how much myelin each part of the brain has,” Sule said of their research, which was funded by the National Institutes of Health.
Specifically, their research approach aims to improve estimates of a patient’s myelin water fraction (MWF), which describes the amount of water within myelin and can indicate brain health. Their method would complement MRI, which is often imperfect at separating myelin from other features in the brain. Sule co-authored a paper showing that a machine learning technique called regularization can help tune out “noise” in the MRI data and improve the accuracy of MWF measurements. “This could serve as an additional diagnostic tool,” Sule said of the potential applications. “For instance, when a radiologist looks at a copy of a brain scan, they could have a tool on their screen that says the estimated myelin content. This could help them figure out what's going on in the brain.”
In separate research with Cameron, Sule develops new ways to speed up simulations of rare molecular events. Many processes that interest researchers—like how proteins fold or how chemicals react—happen too slowly to study on their actual timescales. “It's like trying to see how a glacier melts by observing every centimeter of where the glacier is every week—you’re never going to see it in real life,” Sule said. “A lot of my work with Maria Cameron is about developing new mathematical techniques to understand these rare events.” While Sule focuses on getting the mechanics right, his research could one day advance the search for new and improved drugs to target a variety of human diseases. “One of our target goals is to find what's called the transition rate, which is connected to the rate of a chemical reaction,” Sule said. “It’s super relevant to drug design or studying chemical reactions.”
As Sule sharpens his research skills and reflects on everything he’s learned at UMD so far, he’s surprised by how far he’s come—as a mathematician and as a researcher working on real-world problems.
“Coming to graduate school, I never would have imagined the research I've worked on, especially the applications,” Sule said. “I very much still think of myself as a mathematician, but the exposure I've gotten to medical imaging, drug discovery and just knowing what things people outside of the math world care about has been really exciting and important to learn.”
Written by Emily Nunez
She graduated with a bachelor’s degree in mathematics and a bachelor’s degree in physics.
Jade LeSchack was selected as the student speaker for the University of Maryland’s College of Computer, Mathematical, and Natural Sciences 2025 Undergraduate Commencement Ceremony on May 22, 2025.. The ceremony will honor the college's August 2024, December 2024 and May 2025 graduates receiving bachelor's degrees.
Jade LeSchack is graduating with a bachelor’s degree in mathematics and a bachelor’s degree in physics in May 2025. She is a Design Cultures & Creativity student in the Honors College, and her capstone project, Black Creatives Matter, won an award for creativity in pursuit of anti-racist justice.
LeSchack conducted research in two quantum science groups at UMD. She worked in the Porto-Rolston ultracold atoms lab on experimental projects with electronics and lasers. She currently conducts research in Nicole Yunger Halpern’s quantum steampunk group, studying how thermodynamic laws and phenomena arise in quantum systems. She received a MathQuantum Fellowship from UMD’s Institute for Physical Science and Technology to conduct her research. Beyond UMD, LeSchack was an undergraduate research assistant at the University of Waterloo and studied abroad at the University of Zürich.
LeSchack is active in UMD’s quantum ecosystem and participates in and organizes quantum computing hackathons around the world. She also founded the Undergraduate Quantum Association in her first semester to connect students with UMD’s resources in quantum science and technology. She led numerous initiatives through the club, including the quantum track of the Bitcamp hackathon and an annual quantum career fair, which is now the Quantum Leap Career Nexus. She has been a Society of Physics student member, volunteering as a tutor and physics demonstrator, and a Startup Shell Fellow. She plays for the UMD Women’s Club Ultimate frisbee team and is a member of Omicron Delta Kappa National Leadership Honors Society.
After graduation, LeSchack will pursue her Ph.D. in quantum physics at the University of Southern California in the fall.
At Google Cloud, Valerie Hyde (M.S. ’05, Ph.D. ’07, applied mathematics & statistics, and scientific computation) taps into cutting-edge technologies to take business enterprises into the future.
Whether it’s making sure a virtual agent gets a customer’s drive-thru burger order right or helping an online shopper find that flowered dress they’re looking for, Valerie Hyde (M.S. ’05, Ph.D. ’07, applied mathematics & statistics, and scientific computation) is all about problem-solving for businesses—and doing it with next-generation technologies like artificial intelligence (AI).
For the past five years, as head of the North American Artificial Intelligence Customer Engineering team at Google Cloud, Hyde has tapped into the power of technologies like AI and machine learning to tackle big and small challenges for businesses, drawing from her 20 years of experience at companies including Microsoft, SAS and IBM.
It’s not where she thought statistics would take her 20 years ago—it’s better.
“I never could have imagined this. Sometimes I can't believe that I get to do what I'm doing at this time in history at such an innovative company,” Hyde reflected. “For so many years, it seemed like only a small group of people really cared about statistics and data science, but now everybody is doing AI, and it's really people interacting with technology.”
For as long as Hyde can remember, math and science were always a part of life.
“I was always surrounded by it—both of my parents were professors at state schools in New York—my dad was a chemistry professor, and my mom was a computer science professor,” Hyde explained. “I always liked math the most in school, even more than science, because there were problems that I could solve definitively. That gave me a lot of confidence in my ability to solve problems and think logically.”
As an undergraduate majoring in mathematics and economics at Binghamton University in New York, Hyde discovered that the math classes she liked best focused on statistics. By the time she earned her degree in 1998, she knew that applying math to solve real-world business problems was something she wanted to do.
She landed a job with the antitrust group at the U.S. Department of Justice, where she found herself working on a high-profile case that was getting plenty of attention.
“I used math and programming to help the government quantify its antitrust cases,” she recalled. “The case that people may remember was when the Justice Department sued Microsoft over packaging its Internet Explorer browser with its Windows operating system.”
Though Hyde was working full time, she was still thinking about graduate school—and getting some gentle encouragement from her father.
“Pretty much every call with my dad was, ‘When are you going to go back to school? When are you going to get a higher degree?’ I felt like that was what I was supposed to do,” Hyde recalled. “Both my parents are Ph.D.s, and I think my dad envisioned that I would do that as well. He wanted to make sure that I was always in a stable position, which is possible with advanced degrees in math and science.”
In 2002, Hyde started graduate school at the University of Maryland, not coincidentally, where her father earned his Ph.D. in chemistry in 1969. She soon connected with Mathematics Professor Benjamin Kedem, who quickly became one of her favorite professors.
“Professor Kedem was a mainstay in the Statistics Department. I remember he admitted me into the program because I had an econ background and had solved applied math problems. Another favorite professor was Paul Smith, who taught linear models, which I really enjoyed—it was more applied, doing linear algebra to calculate model parameters,” Hyde said. “I took an independent study with him where we examined advanced statistical and machine learning problems. At the time, using computer vision models to detect handwritten text and numbers was newer; it wasn’t an easily solved problem like it is now.”
Summer internships along the way took Hyde to the U.S. Census Bureau and AT&T.
“It was really neat working at AT&T Research; there were so many smart researchers doing interesting work there,” Hyde recalled. “The project that I remember the most was research on speech-to-text. Even though I didn’t work on that project, I was one of the sample voices used to train their models.”
Challenge after challenge, Hyde worked her way forward.
“I think the biggest thing that I got out of grad school was tenacity. The work was hard for me, and I learned to take it on in small pieces and build accomplishments,” Hyde explained. “I was part of a study group, and we spent the first few years together in one of the rooms on the third floor of the science and engineering library. We each had different strengths and worked together to help get our group through. Sometimes I think I never would have made it through without them and I hope they would say the same thing about me.”
Hyde earned her master’s degree in 2005 along the way to her Ph.D., eventually publishing three papers focusing on price curves, bidding distribution and other aspects of eBay auctions, which were highly popular at the time.
After earning her doctorate in 2007, Hyde spent the next several years putting her statistics, modeling and data science skill set to work in various roles, working in marketing analytics at Accenture, and technical sales building proof of concepts at SAS, IBM and Microsoft before joining Google Cloud in 2020. There, she leads a team of customer engineers at Google Cloud, helping clients incorporate technologies like AI and machine learning into smarter solutions to take their businesses into the future.
“We help our customers understand the technology, incorporate it into their business, and troubleshoot as they implement the solution. Hyde said. “I like working with customers to help them solve a problem in a way that they might not have thought about.”
Many of these business clients find that their customers notice this next-generation problem-solving, like adding AI-powered virtual assistants at fast food restaurants that can understand exactly what someone is ordering.
“That used to be very difficult to solve because it would have been speech-to-text models to intents and then a huge decision-tree like conversational architectures to try to understand what somebody wanted—and it was hit or miss whether it would get it right,” Hyde said. “But now, with large language models, you don't have to do that— the models are able to understand different versions of how you would order your burger.”
For retail businesses whose customers are looking for specific products, AI takes searches beyond keywords to semantic search or even visual search to help people find exactly what they’re looking for.
“In just a few years, search results have improved dramatically with the improvement of multi-model models. You can upload a picture with a flower pattern you like and say with text or voice that you want that on a dress, and AI will understand that you’re searching for a dress that has flowers on it.” Hyde said. “And the thing is, consumers are starting to expect all of these things.”
Though the technology she works with every day is beyond anything she imagined 20 years ago, Hyde still relies on the valuable math and statistics skills she developed at UMD.
“At Maryland, I learned to understand how to solve problems rather than just memorize how to solve problems,” Hyde explained. “I now have the ability to understand and explain what is going on with a speech-to-text model or a computer vision model or a large language model because I have the underlying knowledge of statistics, modeling and the assumptions that you have to make to build the models. UMD really gave me a great foundation.”
For Hyde, it’s all about problem-solving on the cutting edge of technology. And she wouldn’t want to be doing anything else.
“The technology is changing so rapidly that it almost feels old shortly after being introduced because research in AI is flourishing,” she said. “Google is an innovative company with a strong research arm, so they're always at the forefront of what's going on, giving us new ways to solve problems. That’s really exciting.”
Written by Leslie Miller