Mention data, and thoughts may turn immediately to the insatiable data needs of artificial intelligence. But AI and large language models are just one of the forces driving the need for — and generating — large volumes of data. Consider, for instance, the broad field of materials science.

“At the end of the day, it’s the data that matters,” says UC Santa Barbara professor Tresa Pollock. She and McLean Echlin, a research scientist in her lab, know too well that progress in their own area of study and many others depends on instruments and experiments that generate enormous amounts of multimodal data, all of which has to be moved, stored, and retrieved over various time periods. The capability to merge large amounts of data from many different sources is at the heart of new discoveries across many fields.

Facing the recent data explosion ignited by increasingly sophisticated scientific instrumentation across the UCSB campus, Pollock, the Alcoa Distinguished Professor of Materials and then the interim dean of the UCSB College of Engineering (COE), successfully applied for a $250,000 grant from The Hearst Foundation. It enabled the purchase of state-of-the art hardware to accelerate data acquisition, transfer, and analysis and to dramatically enhance storage capacity. That allowed many more graduate and undergraduate students to engage in research in this important emerging area, serving a Hearst Foundation goal of “preparing students to thrive in a global society.” The grant was also instrumental in UCSB’s ability to secure a $5 million NSF grant in 2025 to further advance the university’s Multimodal Imaging initiative.

“One of the foundational capabilities of the Materials Department is the continuously evolving suite of very high-end materials-characterization instruments, including electron microscopes and computed tomography (cT) machines, which now routinely generate large volumes of 3D and 4D (time dependent) data,” Pollock reports. Attracting some five hundred total facility users from departments across campus, she adds, “The instruments in the Materials Department facilities alone can generate tens of terabytes [one thousand gigabytes] worth of data in a few hours of operation.” 

“The typical Google account on campus might accept gigabytes (GB) of data, so cloud computing can serve many research needs,” Echlin adds. “But we can easily generate that by imaging a single, one-micron-thick laser-sectioned slice of a material [during a 3D tomography measurement]. Such huge amounts of data require a bespoke way of dealing with it. The latencies involved in sending and retrieving so much data to and from the cloud — which can be many days — are too long. The cloud can only do so much. In the system made possible by the Hearst grant, we're talking about much faster ethernet connectivity to large storage pools, where no cloud is necessary.”

Data moves so rapidly that speed might not seem an issue, but, when many billions of data bits are being sent or retrieved, latencies lasting milliseconds add up fast. Thus, the closer the site where data is generated is to where it is stored and processed, the better.

Google Says “Bye”

So, how do researchers know when they have too much data and a new approach is needed? One sign, Pollock says, is that Google “kicks you out of the cloud. That’s what they did to us. We had a month to shrink our data or move it all somewhere else.”
 
In giving data-driven researchers the boot in fall 2023, Google identified some “top offenders,” Pollock continues. “We were among them, although I took that as a bit of a point of pride. Echlin was required to reduce his cloud data storage by ninety-five percent.”

Materials professor Daniel Gianola was also flagged as a data “culprit.”

“Our group, in collaboration with several others on campus, performs advanced electron microscopy and diffraction to study structure-property relationships in metallic and ceramic materials used in extreme environments,” Gianola explains. “These instruments are equipped with ultrafast state-of-the-art cameras that are sensitive down to single-electron events and can generate upwards of a terabyte of data per minute, which then must be processed efficiently to decipher the finest details of our materials. The cameras, several of which are located in the core Microscopy and Microanalysis Facility (MMF) on campus, have made the handling of data a central Google-cloud challenge representing a moment of deep data truth.”
 
“The exodus from Google meant that we had to start placing things proximal to each other [to gain the above-mentioned speed advantage],” Echlin says, adding that previous efforts on campus aimed at placing data creation, processing, and storage close to each other had been “mostly serendipitous.” 

Echlin had for some time been seeking to optimize use of the High Performance Computing Center (HPC), which is part of the UCSB Center for Research Computing (CRC) in Elings Hall — also the site of the microscopy suite. “Having the infrastructure for connectivity in one building rather than spread across campus, unifies a lot,” Echlin says. “Connecting those entities was an essential step that gave us places to put data archivally and also to literally stream the data directly from an instrument to storage as we're generating it.”

“Prior to receiving the Hearst grant, users were at the mercy of slow data rates to store, process, and archive data from those instruments, and they always struggled to have effective data backup and archiving options,” Gianola adds. 

“The grant enables streamlined workflows across the data ecosystem. I believe that this system will serve as a model for other shared experimental facilities around the country,” Pollock adds, “Faculty like Dan Gianola, who in March received the 2026 Brimacombe Medal  from The Minerals, Metals & Materials Society for outstanding mid-career scientists, will need these foundational data capabilities to continue to lead the field.”

As a result of the grant, Echlin says, “At least temporarily, the campus data needs are under control. We have a decently large buffer between how much storage we have and how much we need, but we’ll need more in a couple of years.”

System hardware like that installed in the HPC is only as valuable as the technical support provided to maintain it and ensure the integrity of, and access to, data, which HPC provides. “They lease us some real estate and give us some real estate, where we can put some of the servers, and they help us maintain them,” Echlin says. “Support also comes from GRIT (General Research Information Technology), which has the focus of ensuring that the needs of researchers and scientists are represented on campus.” 

A duplicate data setup has been placed in the Chemistry Building and mainly serves  the cryo-electron microscope, which also generates vast volumes of data. Dorit Hanein, a faculty member in the Departments of Bioengineering as well as Chemistry & Biochemistry, manages the facility.

Campus Archival Resources

A key element of improving long-term data storage for all research efforts on campus, derives from a very old technology that has been improved and now occupies a new data-storage niche. Says Pollock, “For archival data, we're moving to high-density tape.”

“Tape is associated with the birth of the computer age, but the densities have gone up, so that, even though it’s slower, a lot of data centers use it to store infrequently accessed data,” Echlin adds. “In our scenario, it will be used for data that was made long enough ago that it probably can go on a shelf, and if somebody really needs it, then they can get it out of that deep storage; they just won’t have immediate access.”
 
The result of the collaboration involving the UCSB Research Office and Hearst resources, an archival tape system has been established as a service to all researchers on campus to ensure long-term access to their data.

The Future of Data

In terms of data’s future, Pollock says, “I don't see any let up; there are valuable discoveries to be made as we develop algorithms to learn from large, multimodal datasets and understand their complexities. We can only make these discoveries if we have the large datasets at our disposal. Looking further ahead, she adds, “As a field, we want a large, up-to-date digital library of materials that we can use to discover, design, and manufacture better materials for the future.” There will undoubtedly be developments in AI that enable us to operate on the data and accelerate the cycle for designing new materials. We are just at the beginning of what is an exciting time to be a materials scientist. The first step, though, is to ensure that the data lives somewhere near where it is generated, and we are grateful to the Hearst Foundation for making that possible.”

Continuing a strong tradition at UC Santa Barbara’s Robert Mehrabian College of Engineering at UC Santa Barbara, Yuheng Bu, assistant professor in the Computer Science Department, has received a prestigious Early CAREER Award from the National Science Foundation (NSF). 

The ability of generative artificial intelligence (AI) to produce text at scale has created an urgent need for trustworthy ways to identify and trace AI-generated content. The project for which Bu received the CAREER Award, titled “LLM Watermarking and Beyond: Foundations and Algorithms via Distributional Information Embedding,” is aimed at advancing watermarking, a family of methods that embed a hidden signal into generated text so that it can be identified later, while maintaining the text’s usefulness and naturalness.

We caught up with Bu earlier this month.

Q: Can you describe generally your intention to develop an attribution model that is more reliable than current approaches?
Yuheng Bu: “The existing practice of watermarking cannot encode more than a single yes-or-no signal, telling us only whether or not a piece of text appears to be watermarked. This binary identifier is useful for basic detection, but is often not sufficient for richer attribution orforensic use, because it cannot reveal which model generated the text, when it was produced, or with whom it was associated.

Q: How does your approach improve on binary watermarking to make LLMs more useful? 
YB: In our approach, metadata, such as the model version, generation source, timestamp, or user-level attribution information, can be encoded. This richer information would make watermarking more useful, since it supports not only detection, but also fine-grained tracing and accountability.

Q: Does your approach address the issues of watermark forgery and erasure? 
YB: Yes. Watermarks can often be removed by rewriting the text without changing its meaning. For example, an LLM can paraphrase the text, or it can be translated into another language and then translated back, which may preserve the meaning while disrupting the watermark signal. This means that the watermark may not survive even when the content itself remains essentially unchanged.
     A related concern is watermark forgery or spoofing. In that case, an attacker may generate harmful content, such as hate speech, that falsely appears to carry the watermark of a legitimate system. This can damage the credibility and reputation of the watermarking scheme, because it creates the impression that the protected system produced content that it actually did not.
     To address these challenges, we need watermarking methods that are both more robust to removal and more secure against forgery. 

Q: How might your research support responsible use of generative AI in research, education, and general society?
YB: One way is by improving tools for protecting intellectual property in datasets, increasing trust in automated reviews and other AI-assisted writing, and supporting secure communication among AI systems. These goals are achievable through our proposed research, in which a general framework would be developed to support these applications. At the same time, full real-world impact will also require follow-on work and broader efforts in the same direction.

Q: Speaking of the future of AI research, can you tell us about the educational component of the project?
YB: Educational activities include developing hands-on training via short course modules for high school students, interactive workshops for junior high families, and workshops for high school science teachers. Undergraduate and graduate students will have opportunities for rich learning embedded in competitions around watermarking to increase AI security. These efforts will give students early practice in thinking about reliability, security, and design trade-offs in generative AI, so that they learn to build and evaluate AI systems responsibly from the start.

Q: The proposal mentions embedding multi-bit information into text generated by LLMs. Why is that so challenging?
YB: Embedding multi-bit information means encoding a small amount of metadata into the generated text. It’s challenging because text has limited freedom: the model must still produce fluent, natural, and semantically accurate language. The more information we try to embed, the harder it is to preserve text quality while also maintaining robustness and security.

Q: Can you do it?
YB: We are making progress toward this goal. We have developed a theoretical framework for analyzing text quality, robustness, and security in zero-bit watermarking, and we are exploring different strategies to generalize this framework to multi-bit watermarking. Currently, we can embed a few bits in a sentence, but we believe there is room for improvement.

Q: What is the distributional information embedding problem mentioned in the proposal? 
YB: In simple terms, traditional watermarking often works by inserting identifying information into an existing text passage. In contrast, for generative AI watermarking, the information is embedded during the creation process itself. That means that we gently steer how the model generates content by adjusting the probability distribution over next-token predictions so that the final output carries a hidden signature that can later be detected.

Q: What is meant by the sequential generation of LLM–generated text?
YB: Generated text is sequential because each new word depends on the words that came before it. For example, after “The cat sat on the,” the next word is much more likely to be “mat” than something unrelated. That is very different from independent samples, where each draw is made separately and does not depend on previous ones, like repeatedly drawing numbers from — and returning them to — a bag. So, “dependence” here means context dependence across tokens, or small pieces of text that an LLM can generate.

Q: Can you talk a little about the tradeoff, mentioned in the proposal, between watermark robustness and preserving natural text?
YB: One key trade-off is between information rate and text quality. That is, the more information we try to embed in the text, the harder it is to keep the output fully natural and fluent. Another is between robustness and detectability, because while making the watermark stronger can improve detection and make removal harder, it may also increase distortion or make the pattern easier for an attacker to identify and spoof.

Q: This project proposal has an ambitious list of goals. Can you realize all of them? 
YB: A single project is unlikely to fully solve the problems of watermark forgery and removal, so our contribution is best understood as what we hope will be a meaningful foundation, not a complete solution. The goal is to develop algorithms and authentication mechanisms that substantially improve our ability to distinguish genuine watermarks from forgeries and to identify likely removal attempts, with provable guarantees in well-defined settings.

Q: What makes “in-context” watermarking different from its predecessors?
YB: Unlike most existing watermarking methods, which require access to the model’s decoding process, in-context watermarking embeds the watermark through the user prompt alone, using the LLM’s in-context learning and instruction-following ability.
     This makes it different in two ways. First, it is model-agnostic: the party applying the watermark does not need control over the model internals or the decoding algorithm. Second, it is especially useful in settings such as AI-generated peer reviews, where organizers may suspect LLM use but have no access to the underlying model. In that case, the watermark can be induced through carefully designed prompts, and later detected from the generated text.

Nearly thirty current students, incoming graduate students, and recent alumni from The Robert Mehrabian College of Engineering (COE) at UC Santa Barbara have been awarded prestigious graduate research fellowships from the National Science Foundation (NSF), recognizing their exceptional promise in science and engineering. The NSF awarded more than 2,500 fellowships for the 2026-27 academic year from a pool of nearly 14,000 applicants. Recipients are selected based on intellectual merit and broader impacts, including their potential to advance scientific discovery and contribute to society.

“The NSF GRFP is one of the clearest indicators of future leadership in science and engineering,” said Umesh Mishra, dean of The Robert Mehrabian College of Engineering. “Seeing so many of our students recognized at this level speaks to the culture of innovation, rigor, and collaboration that drives discovery at UC Santa Barbara, both at the graduate and undergraduate levels, and impacts the global economy.”

The NSF Graduate Research Fellowship Program (GRFP), one of the nation’s most competitive honors for graduate students in STEM fields, provides three years of financial support over a five-year period. Each fellowship includes a $37,000 annual stipend and a $16,000 cost-of-education allowance, totaling $159,000. 

This year’s recipients from COE include fifteen current students, at least eight incoming PhD students, and six alumni who are now pursuing graduate degrees at other institutions. The 2026-27 cohort represents a wide range of disciplines and research areas across the college, spanning six departments from COE: bioengineering, chemical engineering, computer science, electrical and computer engineering, materials, and mechanical engineering. 

2026-27 GRFP Recipients from The Robert Mehrabian College of Engineering

Current UCSB students (14)

Ethan Chen, Materials PhD student

First-year PhD student, Ethan Chen studies the complex supercurrent behavior of hybrid Josephson junctions based on thin-film cadmium arsenide, a two-dimensional topological insulator. His research is aimed at advancing fault-tolerant qubits, a key step toward more efficient quantum computing systems.

Advised by materials professor Susanne Stemmer, Chen says the NSF Graduate Research Fellowship provides both validation and opportunity. “The fellowship affirms the confidence I have in my scientific proficiency and highlights the impact of my research support system,” he said, adding that it enables him to pursue more curiosity-driven research into quantum materials.

Keyes Eames, Materials PhD student 

Advised by Distinguished Professor Steven DenBaars, Keyes Eames is a first-year PhD student broadly focused on gallium nitride (GaN) optoelectronics, with potential applications ranging from energy-efficient data communication to ultraviolet sterilization technologies.

For Eames, the fellowship’s greatest impact is the flexibility it provides. “The most important professional impact of the award will be academic freedom,” he said. “The flexibility enables me to choose the most impactful research and career path. On the personal side, the incredible research opportunity and implied career trajectory provided by this fellowship feel like an incredibly exciting and weighty responsibility.”

Trevor Hagan, Materials PhD student 

First-year PhD student Trevor Hagan, who is co-advised by Rachel Segalman and Craig Hawker, is developing electrostatically crosslinked polymer networks with applications in plastic waste recycling. His research focuses on enabling more efficient reuse and repurposing of polymers, advancing sustainable materials solutions.

Raised in rural southern Indiana, Hagan credits his parents for nurturing his early interest in science, often traveling long distances to support his curiosity. Receiving the fellowship marks both a personal and academic milestone.

“I am the first in my family ever to attain an academic degree, and thus also the first to pursue a PhD,” he said. “Having received an NSF Fellowship is both personally and academically affirming that multiple blind reviewers found my background, training, and research ideas compelling enough to invest in.”

Anika Jena, Chemical Engineering undergraduate 

Anika Mahajan Jena, a chemical engineering undergraduate senior, will pursue her PhD at Stanford University, where she will develop and probe advanced functional polymeric materials for health, sustainability, and human advancement. As an undergraduate researcher at UCSB, Jena studied phase separating membrane-actin network composites and tuned their mechanical properties with then-chemical engineering assistant professor Sho Takatori. Later, she engineered supramolecular architectures by coupling DNA nanotubes to condensates with physics professor Deborah Fygenson.

Jena, a 2024 Congressional Goldwater Scholar, explains that the NSF Fellowship will support her research beyond traditional funding constraints. “This fellowship will enable me to freely conduct independent, cross-disciplinary work on self-driven research projects,” she said. She indicates possible future research in self-healing, stimuli-responsive, conductive, and biocompatible soft materials.

Christopher Koh, Mechanical Engineering PhD student 

A first-year PhD student, Christopher Koh is exploring the intersection of control theory and machine learning to advance safer, more reliable intelligent systems. Advised by Francesco Bullo, his work focuses on integrating the mathematical rigor of control theory with the flexibility of machine learning, an approach aimed at improving performance in safety-critical applications such as robotics, aerospace systems, and power grids.

While machine learning offers powerful new capabilities, its lack of reliability in high-stakes environments remains a key limitation. Koh’s research seeks to address that challenge, developing methods that can meet the demands of real-world deployment where precision and dependability are essential.

Receiving the prestigious NSF Fellowship provides both validation and stability early in his graduate career. “Securing funding in the current environment at the federal level provides a large sense of relief,” Koh said. “This allows me to focus more fully on my research and degree requirements, and to pursue work that aims to make these systems more flexible, safe, and reliable.”

Ben Kunimoto, Bioengineering PhD student; Data-Driven Biology Trainee

Ben Kunimoto, a first-year bioengineering PhD student advised by Siddharth Dey, is developing a new technology to map the subcellular localization of the transcriptome, enabling researchers to determine where mRNA from all genes is situated within individual cells. His work addresses a longstanding challenge in biology: while nonuniform mRNA localization is known to play a critical role in processes ranging from embryogenesis and tissue development to immunology and neuroscience, it has been difficult to study at a genome-wide scale due to technological limitations. Kunimoto aims to overcome this barrier and apply the method to investigate how polarized mRNA localization influences cell differentiation during early mammalian embryogenesis.

“I feel very honored to have received this fellowship, and I’m thankful that it will help me and my lab pursue interesting and valuable research,” he said.

Karlee Macaw, Mechanical Engineering master’s student 

Karlee Macaw, a first-year master’s student advised by Ryan Stowers, studies how cells respond to the mechanical properties of their environment, with implications for diseases such as cancer and fibrosis.

Macaw describes the fellowship as a pivotal milestone in her academic journey. “It validates the work I’ve committed to and makes pursuing a PhD focused on mechanobiology research a real possibility,” she said, noting that it provides the freedom to focus on impactful research. 

Conor Puglsey, Chemistry & Biochemistry Undergraduate student

A fourth-year undergraduate triple-majoring in chemistry and biochemistry, pharmacology, and statistics and data science, Conor Pugsley has conducted research in the Materials Department since his first day on campus. Working with materials associate professor Angela Pitenis, he has investigated how polymer architecture influences mechanical performance, with the goal of designing more effective materials for biomedical implants. A Beckman Scholar and active participant in Center for Science and Engineering Partnerships (CSEP), Pugsley has built a strong foundation at the intersection of chemistry, materials science, and data-driven research.
 
He will pursue a PhD in Bioengineering, where he plans to study the chemistry of biomaterials to enable precision drug delivery of complex macromolecular therapeutics. His proposed research focuses on using supramolecular chemistry to control hydrogel mesh structures, allowing for more precise encapsulation and release of therapeutics—an approach with potential to reduce overdose risks and improve automated dosing in medical treatments.

“Receiving this fellowship reflects positively on my drive to pursue research and highlights the excellent mentorship I have received along the way,” Pugsley said. “It enables me to pursue my own work in graduate school and focus on the intellectual curiosity that initially brought me to biomaterials research, while laying the foundation for securing future research support as I work toward becoming a professor.”

Naomi Rehman, Computer Science PhD student 

A first-year PhD student advised by Tim Sherwood and Jonathan Balkind, Naomi Rehman works at the intersection of computer architecture and artificial intelligence, aiming to improve AI efficiency and enable privacy-preserving systems on edge devices.

Rehman says that the fellowship affirms her academic path and opens new possibilities. “When I switched from robotics to computer engineering with the hope of becoming a computer architect, I was nervous about whether it was a wise decision. Receiving this fellowship feels like a confirmation that it was,” she said. “I’m especially proud to represent women in computer architecture, and I hope receiving this encourages other women to pursue careers in the field. Academically, I’m extremely grateful for the freedom that this fellowship gives me to pursue the work that I find most promising and impactful.”

Ava Salami, Chemistry undergraduate 

A fourth-year chemistry and biochemistry major, Ava Salami works as an undergraduate researcher in the lab of bioengineering assistant professor Marley Dewey. She investigates the use of cell type-specific extracellular vesicles within biomaterials toward tissue-specific repair in the musculoskeletal system. 

Rohil Shah, Computer Science undergraduate

A fourth-year undergraduate in the College of Creative Studies, Rohil Shah studies computing with a focus on search systems and artificial intelligence. His research centers on designing efficient, time-aware search engines that can rapidly adapt to new or evolving information, with applications ranging from real-time information retrieval to combating misinformation during breaking events. After graduating, Shah will pursue a master’s degree in computer science at Stanford University.

Shah credits his mentors—especially computer science professor Tao Yang, with whom he works most closely—and the broader UCSB community for shaping his academic path and research direction.

“Receiving the fellowship feels like the product of their mentorship and guidance,” he said. “It motivates me to take full advantage of this opportunity and to explore the intersection of search systems and artificial intelligence in order to develop technologies that can meaningfully improve people’s lives.”

Roenigk Straub, Materials PhD student

Roenigk Straub, a first-year PhD student advised by Daniel Oropeza, studies parameter-induced porosity in additively manufactured metals, with specific applications in aerospace, electronics cooling, and catalysis.

The fellowship has increased Straub’s confidence as he transitions into academic research. “Prior to coming to UCSB, I had been focused on working in industry, so receiving this fellowship has increased my belief in myself as a researcher,” Straub said. “Learning that others see potential in my work further motivates me to make the scientific breakthroughs I outlined in my proposal a reality.”

Straub says the fellowship will allow him to broaden the number of alloys and functional properties that he can examine, expanding the potential impact of his work to a wider range of next-generation technologies.

Andres Torres, Mechanical Engineering PhD student 

A first-year PhD student advised by Elliot Hawkes and Tobia Marcucci, Andres Torres is developing autonomous capabilities for soft robots designed to safely interact with humans, similar, Torres says, to “Baymax from the movie, Big Hero 6.” 

Torres describes the award as both a personal milestone and a reflection of strong mentorship at UCSB. “Receiving the NSF Fellowship means so much to me,” said Torres. “I definitely would not have won this without the amazing resources and faculty here at UCSB, especially my advisors. This fellowship will enable me to pursue my dream of working on robotics research!”

Christopher Xu, Mechanical Engineering PhD student 

Christopher Xu, a first-year PhD student advised by Elliot Hawkes, develops highly agile robotic systems capable of navigating complex environments, with applications in exploration and environmental monitoring. For example, he is working on a robot that jumps into trees like a squirrel.

Xu says the fellowship allows him to fully immerse himself in research. “I can focus entirely on research, which is what I came here to do,” he said, adding that the support gives him the freedom to explore ambitious ideas that might otherwise be difficult to pursue. “I love the work that I do in the lab like a hobby, and I am grateful to be able to do it with so much flexibility.”

Roland Yin, Electrical Engineering undergraduate student

Roland Yin, a senior undergraduate in electrical engineering, conducts research on quantum and inorganic materials under the guidance of professors Stephen Wilson and Ram Seshadri. His work explores structure-property relationships, beginning with energy-efficient synthesis of sodium battery cathodes and evolving to superconducting materials for quantum applications.

Currently, Yin investigates electronic correlations in kagome superconductors, a quasi-two-dimensional class of materials that exhibit an interplay between unconventional superconductivity and charge-density waves. By selectively alloying and doping these systems, he aims to understand how electron–electron interactions emerge and to leverage intrinsic Josephson effects for quantum sensing applications, including high-sensitivity radio-frequency detection in quantum computers and atomic clocks.

“Receiving the NSF GRFP is both exciting and deeply validating,” Yin said. “It recognizes the research I have pursued over the past few years as a meaningful contribution to quantum technologies, and it affirms that I have the potential to grow into an independent researcher who can frame interesting questions, discover innovative solutions, and communicate noteworthy results. As someone who envisions a future in academia, being selected by NSF gives me a strong sense of support heading into my PhD and academic career.”

In addition to the fourteen current engineering students, at least eight incoming PhD students who will begin their studies at UCSB this fall, and six recent alumni now pursuing graduate degrees at other universities, also received NSF Graduate Research Fellowships. 

Recent UCSB alumni now pursuing graduate degrees at other universities (6)

• Stephanie Anujarerat (Chemical Engineering), now at Carnegie Mellon University

• Marcus Condarcure (Chemical Engineering), now at Purdue University

• August Dolmatch (Chemistry), research assistant in the lab of Carolyn Mills (Bioengineering)

• Aaron Huang (Physics), undergraduate researcher in the lab of Omar Saleh (Materials), now at the University of Chicago 

• Sofia Rivalta Popescu (Chemical Engineering), now at Stanford University

• Anton Semerdjiev (Chemical Engineering), now at the California Institute of Technology

Incoming PhD students who will begin their studies at UCSB in fall 2026 (8)

• Three in the Electrical & Computer Engineering Department

• Two each in the departments of Bioengineering and Materials 

• One in the Chemical Engineering Department

Established in 1952, the GRFP has supported over 70,000 graduate research fellows, many of whom have gone on to become leaders in research and innovation. More than 40 former fellows have received Nobel Prizes, underscoring the program’s long-standing role in advancing scientific innovation and leadership. 

“It’s one of those must-do classes. Honestly, there's not going to be a better opportunity to really dig into a specific project and see it to fruition.” That’s how fourth-year computer science student Cooper Hawley describes the UC Santa Barbara Computer Science (CS) Department’s two-quarter capstone course sequence, which brings together student teams and industry leaders to tackle real-world problems. After spending six months working on challenges from supporting stroke survivors through a smartwatch app to allowing property managers to easily schedule repair and maintenance work, eight student teams presented their projects in March at the annual CS capstone day.

The event draws undergraduate and graduate students, faculty, alumni, and industry partners for the presentations, a poster session, awards, and networking opportunities for the forty students participating in the capstone project. At the event on March 9, the teams presented their projects in front of a judging committee made up of Chris Bunch from Google, Wei-Tsung Lin from Salesforce, and Alexis Cole from Amazon. 

“Over the years, I've seen the capstone projects evolve from isolated technical prototypes into highly sophisticated, production-ready systems. There is a noticeable trend toward solving high-stakes, real-world problems,” said Chandra Krintz, computer science professor and associate dean of the UCSB Graduate Division. “The sophistication has increased because students are no longer just coding,” noted Krintz, who taught this year’s winter-quarter capstone project course. “They are acting as systems architects, integrating complex distributed agents, leveraging recent advances in artificial intelligence and machine learning, and incorporating advanced cloud services.”

This year, three teams took home awards. First-prize winner Team Altera created one of those sophisticated, production-ready designs, Krintz said. “They tackled the notorious bottleneck of healthcare referral coordination by transforming the multiday medical-referral process into a pipeline that allows patients to schedule referral appointments in less than an hour.”

Team Cadense took second place for the smartwatch-based therapy system they developed to support stroke survivors as they retrain their walking cadence. And third-place winner Team RapidRecall, sponsored by Cottage Health, designed and built a system that automates the analysis of recalls from the FDA and industry suppliers, accelerating a hospital’s ability to intercept harmful medical devices and pharmaceuticals before they reach the patient.

“The judges have repeatedly told us, ‘This is the best day of our year.’ And we have senior people in the industry who see a lot of cool things,” said computer science professor Dahlia Malkhi. She said that, while she knew it was a cliché, the capstone students were all winners. “They all worked so hard, they had phenomenal achievements, and they went from being forty individual students in the fall to being eight cohesive teams, each with a mini company or mini startup in a completely different domain. And their presentations were just staggering.”

Students began the capstone project in Malkhi’s fall-quarter course. (A capstone project in either computer science or computer engineering is a graduation requirement for computer engineering majors, and an optional elective for computer science majors.) In the first few weeks, students formed groups, then heard pitches about potential projects from industry partners AppFolio, Artera, Cadense, Cottage Health, Mysten Labs, Planet Labs, SpaceComputer, and Visual Layer. Groups ranked their preferences, and Malkhi and her team matched the groups with projects. 

From there, students met both in class and then with a mentor as they conceptualized and developed their approach to the problem. Rithwik Kerur, a second-year computer science PhD student and the capstone teaching assistant, also supported the teams from project selection to their final presentations.

Faster, More Accurate Medical Referrals

Hawley and his teammates — fourth-year computer engineering majors Aden Jo, Benjamin Soo, David Duenas, and John Hagedorn — worked with sponsor Artera, which provides a platform used by healthcare organizationsto improve patient access, engagement, and outcomes. With the help of Kerur and Anav Sanghvi, a senior software engineer at Artera, they designed a process that allows doctors and patients to set up and schedule referrals with the help of an AI agent — a process that typically takes several days, but that the team shortened to less than an hour.

Doctors and other health-care providers have access to their patients’ intake interviews and their progress throughout the system, Hawley noted, so that there is human input and oversight at every step.  

Hawley said that the change in his team’s project between the two quarters was unmistakable. “In the first quarter, it felt like it was not coming together. But by the time we entered the second quarter, with the break between the two, we could look back more transparently and see that we’d set a really nice foundation.”

Keeping Movement on Track for Stroke Survivors

Team Cadense — fourth-year computer engineering students Vincent Cheong, Jim Wang, Christopher Lai, Scott Ricardo Figueroa-Weston, and Jeremiah Wong — was inspired by a stroke survivor named Kevin and his efforts to relearn how to walk.  While closely supervised in physical therapy clinics, stroke survivors and others with movement disorders practice movements so that the brain re-learns through practice, repetition, and feedback. But at home, these exercises can be more challenging to maintain, which can affect a patient’s progress.

To address this problem, Team Cadense developed both a device and an app that helps patients continue to train at home. The device attaches to a walking stick and provides an auditory beat that allows patients to sync their movements to visual, audio, and haptic cues. The smartwatch app, which can be used independently or with the device, gives patients feedback as they walk. 

“At first, when we were designing the app, we designed with ourselves — students — in mind,” Figueroa-Weston said. “But when we gave the app to the patients, they didn’t know what to do.”  They used this feedback to create a user-friendly interface with a simple design and large buttons that would be easy to use for older patients who have  physical impairments.

“The most rewarding part of this process was being able to talk with the patients,” Figueroa-Weston said. “That was something that I definitely would not have gotten outside of the class. And it has shaped my process of how I would build things in the future. Now, I know it’s important to talk to your users as soon as possible.”

Capstone Projects Kickstart Careers

While some projects wrap up once the course ends, others continue to grow. Three of the eight capstone projects from last year went live in some form, Malkhi said, including one team that formed a startup that was subsequently funded by the City of Santa Barbara’s incubation program. 

In other cases, student projects become the seeds for further academic research. “I’ve already admitted several students who want to continue their projects,” Malkhi said, “so we’re going to incorporate them into the research we have in my lab.”

Malkhi noted that no matter the outcome of the capstone project itself, the experience has long-term ripple effects. “Some of the students get jobs from the companies they’ve interacted with. Some get jobs from the judges or spectators that came to the capstone event,” she said. “And all the students that come back to talk with me say that in their job interviews, they have a story to tell beyond their transcripts. They can talk about the project and the teamwork, the challenges they’ve encountered, and the full-fledged design and product. They tell me that this is the one thing that gets them job offers, and that they also feel good about interviewing, because they have something to show for themselves.”