Postdoctoral Fellows
Data Science Fellows

Vladislav Ayzenberg
Department of Psychology
Vlad is a developmental cognitive neuroscientist who is interested in understanding the mechanisms that support early developing perceptual abilities in human infants. Because measuring the underlying processes of the infant brain is incredibly challenging, Vlad uses biologically plausible computational models to explore what kinds of processes and developmental constraints may be sufficient to support infant perception. As a postdoctoral fellow with Dr. Michael Arcaro, he is starting to explore how early developing anatomical structures in the neonate visual system may scaffold perceptual abilities later in life.

Chang-Yu Chang
Department of Biology
Chang-Yu Chang is a systems biologist who combines computational and experimental approaches to study the ecology and evolution of microbial communities. His past works focused on engineering the emergent properties of microbial consortia using directed evolution. With Dr. Corlett Wood, he is starting to work on host-associated microbial ecosystems and asks how the interplay between ecological interactions and genomic structures shapes the host phenotypes. Chang-Yu is originally from Taiwan.

Sam Dillavou
Department of Physics and Astronomy
Sam Dillavou is a physicist interested in the overlap between physics and machine learning, and how the two fields can inform and support each other. His projects include building physical systems that can perform machine learning tasks (learn) without a processor, studying complex systems like granular flows that have resisted understanding using standard statistical methods, and using machine learning to make experimental science easier and more accessible.

Andrew Koepp
Department of Psychology
Andrew is an applied developmental psychologist studying how young children learn to manage their attention and behavior in the classroom. He uses wearable accelerometers to capture children’s inattentive and impulsive behaviors, leveraging the objective and continuous recording of movement to understand daily fluctuations in children’s behavior. With support from Penn’s Data Driven Discovery Initiative, Andrew will apply supervised machine learning methods to evaluate which features of children’s movement (i.e., amount, intensity, onset, duration) best predict teacher ratings of children’s impulsivity and inattention. Andrew also looks forward to using unsupervised clustering techniques to identify temporal patterns of children’s behavior and network analysis to study social dynamics in the classroom.

Sarah Lee
Department of Linguistics
Sarah Hye-yeon Lee is a linguist who is interested in the interface between language and cognition. Her work takes an experimental, data-driven approach to understanding the relationship between language and non-linguistic conceptual structure, as well as to understanding the cognitive mechanisms that underlie real-time language processing. Her research uses a range of behavioral data (e.g. reaction time data, eye-tracking data) and corpus data. Sarah holds a Ph.D. in Linguistics from the University of Southern California.

Adrien Thob
Department of Physics and Astronomy
Adrien is an astronomer interested in understanding the mechanisms that drove the formation of galaxies throughout the history of the Universe. Thanks to his multidisciplinary career with a background in telecommunication and data processing engineering, he routinely uses data science, high performance computing and software development in his research. Adrien holds a PhD in Computational Cosmology from the Astrophysics Research Institute of Liverpool John Moores University in the UK, and joined the University of Pennsylvania as a postdoctoral researcher after a first postdoctoral position at the University of Washington.

Carolina Torreblanca
Department of Political Science

Coby Viner
Department of Biology
Coby is a computational biologist interested in how evolutionary processes shape molecular systems, especially through the lens of algorithm design. Cross-trained in computer science and biology, he studied Bioinformatics as an undergraduate before completing his PhD in Computer Science at the University of Toronto. His research has addressed questions in transcription factor binding and DNA modifications such as methylation, using statistical, combinatoric, and information-theoretic approaches. He previously developed Veridical, a computational method for validating mRNA splicing mutations. At Penn, he is exploring how natural selection acts across high-dimensional biological systems, using simulations to connect evolutionary theory with empirical data.

Xiaozhi (Taro) Yang
Department of Psychology
AI x Science Fellows

Noëmi Aepli
Department of Linguistics
Noëmi is a computational linguist driven by a deep fascination with dialects, language varieties, and low-resource languages. Her research addresses the challenges of linguistic diversity, particularly in non-standardized dialects. Her dissertation explored how text-based computational models can better handle linguistic variation. Her current work focuses on speech-based language models to advance NLP in diverse linguistic contexts and gain insights into how linguistic information is encoded in neural networks.

Yahav Bechavod
Department of Computer and Information Systems
Yahav is a computer scientist interested in developing principled techniques to ensure reliable machine learning for consequential decision-making. To this end, he designs algorithms that aim to guarantee fairness in prediction, be incentive-aware, and address bias along the different parts of the ML training pipeline — from data collection, to the feedback structure, to incorporating constraints and characterizing their costs and impacts. His work draws on and combines techniques and concepts from machine learning theory, optimization, and algorithmic game theory. He holds a PhD from the Hebrew University, during which he was also an Apple Scholar in AI/ML.

Sourav Dey
Department of Chemistry
Sourav Dey is a computational chemist working in the intersection of chemistry and machine learning. His current interest is to automate chemical reaction discovery and contribute towards the development of self-driving labs. He uses principles of theoretical chemistry and machine learning to understand molecular properties and their interaction with each other in complex environments. His prior work involves electronic structure theory and multiscale modeling.

Marcelo Guzmán
Department of Physics & Astronomy
Marcelo Guzmán is a physicist working at the frontier of machine learning and soft-matter physics. He studies physical networks as new platforms for the emergence of learning and adaptation. By combining theory and simulations, he takes a theoretical perspective on the fundamental differences between machine learning and physical learning, such as interpretability, scaling laws, efficiency, and robustness. His prior work involves the characterization and design of physical networks with topologically protected responses.

Nicolò Dal Fabbro
Department of Electrical and Systems Engineering
Nicolò is an engineer interested in designing and analyzing algorithms for multi-agent intelligence. With a background in communication engineering and communication-efficient distributed machine learning, he focuses on developing the next generation of autonomous systems in which multiple agents (e.g., robots and autonomous vehicles) will be trained to coordinate and function in the physical world. Recent projects focus on algorithm design for cooperative and communication-efficient multi-agent reinforcement learning, algorithms for coordinating multiple autonomous agents in underwater robotics (for scientific discovery and environmental monitoring), and the use of language to automate and speed up multi-agent cooperation learning.

Xinquan Huang
Department of Mechanical Engineering and Applied Mechanics
Xinquan Huang is currently working with Prof. Paris Perdikaris and Prof. Nat Trask on generative models in science and engineering. His research interests span the areas of physics-informed machine learning, operator learning, generative modeling using diffusion models and their applications to fluid simulation, uncertainty quantification, and inverse problems. He completed his Ph.D. at King Abdullah University of Science and Technology and has interned at Microsoft Research AI4Science.

Hancheng Min
Department of Electrical and Systems Engineering
Hancheng Min is a machine learning theorist whose research centers around building mathematical principles that facilitate the interplay between machine learning and dynamical systems, working with Prof. René Vidal. His recent research focus has been on understanding the inductive bias of the training algorithms on promoting certain structural properties in the neural networks and connecting these theoretical findings to practical issues in ML such as the adversarial robustness of neural networks.

Kieran Murphy
Department of Bioengineering; Physics and Astronomy
Kieran is a physicist entranced by information theory, leveraging machine learning to track down information in general relationships through data. Recent projects have focused on developing methods to localize information: deconstructing complex systems into intelligible approximations — including the Mona Lisa and a simulated glass under shear — and switching around a standard representation learning method to isolate information out of mere groupings of data.

Melanie Segado
Department of Bioengineering
Melanie Segado is a postdoctoral fellow in Bioengineering advised by Konrad Kording at University of Pennsylvania. Her research is currently focused on building computational models of complex human movement using data-driven artificial intelligence approaches. Specifically, she is developing foundation movement models that can be fine-tuned for specialized applications in human health and performance, including low-cost video-based diagnostics for infant movement disorders. She holds a PhD in Neuroscience from McGill University where she studied sensorimotor integration during skilled motor execution. Following her PhD she worked as a research scientist at the National Research Council Canada, integrating neural interfaces into custom-developed virtual reality environments.

Brynn Sherman
Department of Psychology
Brynn Sherman is a cognitive neuroscientist interested in human memory. In particular, she combines insights from human behavior, data-driven analysis of neuroimaging data, and computational models to understand how we use our memories to adaptively guide future behavior. In her current projects, she is taking empirical and computational approaches to understanding how sleep facilitates the integration of new memories with existing knowledge.

Yan Sun
Department of Statistics and Data Science
Yan Sun is a statistician interested in developing trustworthy machine learning (ML) models. His research focuses on uncertainty quantification and model reliability. He develops rigorous methods to evaluate and improve model calibration, enhancing the accuracy of probabilistic forecasts. His work in sparse neural network models identifies key predictive features, making ML models more interpretable and efficient.
AI x Science x Medicine Fellows

Tess Cherlin
Department of Pathology and Laboratory Medicine
Tess is a computational biologist interested in understanding the complex factors underlying diseases. Her research uses data-driven methods to improve disease prediction, detection, and treatment by integrating multiomic data. Some of her current projects in the Setia-Verma lab investigate diseases affecting women across their lifespan like polycystic ovary syndrome (PCOS). In addition to her research, Tess is passionate about teaching and is a PennPORT scholar. Tess earned her PhD in Genetics, Genomics, and Cancer Biology from Jefferson University.

Hayoung Donnelly
Department of Psychiatry
Hayoung Donnelly is a computational psychologist studying how adolescents and young adults at risk for suicide disclose their struggles, and how NLP and LLMs can be useful to understand and amplify their voices. She enjoys discovering and integrating diverse data sources (e.g., social media posts, clinical notes, research articles, and census data) to inform socio-ecological suicide prevention strategies, with a particular focus on youth in South Korea, the U.S., and low- and middle-income countries.

Wei Li
Department of Biostatistics and Epidemiology
Wei Li is a postdoctoral fellow at the Perelman School of Medicine, working with Prof. Mingyao Li. His research focuses on developing computational methods, particularly graph neural network-based models, to analyze complex spatially resolved data and address challenges related to data quality and integration. By developing advanced computational techniques, his work refines biological signal extraction and enables more precise characterization of cellular interactions within tissue architecture.

Sydney Pugh
Department of Biostatistics, Epidemiology and Informatics
Sydney Pugh is a computer scientist interested in building trustworthy and interpretable artificial intelligence (AI) for clinical decision support. Her current research focuses on developing a real-time multimodal AI-based system for the early detection of cognitive impairment, Alzheimer’s dementia, and related dementias. She is leveraging large language models (LLMs) to extract diagnostic clues of cognitive impairment – such as repetitive speech or word-finding difficulty – from patient-provider conversations in primary care settings to inform an interpretable predictive model. Sydney earned her Ph.D. from the University of Pennsylvania where she worked on weakly-supervised methods for evaluating the performance of medical AI systems.

Sheng Wang
Department of Radiology
Sheng Wang’s research integrates human behavioral data with artificial intelligence to advance diagnostic imaging analysis. His past work involved pioneering the use of eye-tracking to enhance computer-aided diagnosis and developing foundational models that combine large language models with medical imaging for interactive and interpretable diagnostic tools. Currently, his research leverages human behavior insights to further improve computational pathology foundation models, optimizing the analysis of Whole Slide Images (WSIs) by significantly reducing processing time and enhancing diagnostic accuracy. Sheng received his Ph.D. from Shanghai Jiao Tong University and is now a postdoctoral researcher in the Department of Pathology & Laboratory Medicine at the Perelman School of Medicine.
Interested in becoming a postdoctoral fellow?
If data science is a significant part of your research and you are interested in cross-disciplinary interactions, you are invited to apply to become a Fellow of our program. Please visit our application instructions page for detailed instructions.
For more questions send us an email.
Former Fellows
Please visit our Former Fellows page to learn more about what our past fellows are up to now.
Data Science Lunch Series
Each semester, our postdoctoral fellows meet most Tuesdays for an informal lunch with a faculty guest speaker to discuss research themes in data science. To learn more about our data science lunch series, please see here.