By Sophia Chen and Thomas Jilk

Assistant Professor Joshua Cape has received a National Science Foundation (NSF) CAREER Award, one of the agency’s most prestigious honors for early-career faculty. Cape joins fellow Statistics faculty members Jessi Cisewski-Kehe and Nicolás García Trillos as recent CAREER awardees.

In his research, Cape develops statistical methods that help researchers make sense of complex data, with applications in fields like neuroscience, psychiatry, and electrical engineering. One line of his work focuses on brain networks associated with schizophrenia, aiming to develop methods that could support more personalized mental health treatments.

With the CAREER Award, which includes a minimum of $400,000 in funding over five years, Cape and his collaborators will develop new tools for creating data embeddings: numerical representations of text, images, video, audio, or graphs that capture meaning and relationships in a way computers can understand. Data embeddings are the building blocks that run “under the hood” of today’s large language models, as Cape puts it. The ultimate goal, as the award’s abstract explains, is “to enable the next generation of effective, user-friendly, scalable, interpretable data analysis pipelines.”

Below, Cape discusses what the award means to him, the innovative research it will support, and his simple but powerful advice for students and early-career researchers.

What does receiving the NSF CAREER Award mean to you?

I’m incredibly grateful to everyone who helped make this NSF CAREER Award possible – the NSF program officers and reviewers, my collaborators, my mentees, and my former mentors.

This award represents an investment in the ongoing development of statistical methods for extracting meaning and actionable insights from complex datasets arising in the natural sciences and in engineering.

Your CAREER Award supports work on developing new statistical learning and data‑embedding tools. Why is developing these tools important?

As a discipline, Statistics can be understood as the “science” in “data science.” Embeddings, which are representations of data or of concepts more generally, are an important building block for many statistical procedures, machine learning algorithms, and artificial intelligence systems.

Constructing and working with embeddings is broadly useful and applicable, not just for numerical tabular datasets but also for text, images, and even audio recordings. In fact, embeddings are routinely used under the hood in large language models such as Google’s Gemini and OpenAI’s GPT-5.

In this project, we seek to study the statistical foundations associated with embeddings using tools from probability theory, high-dimensional statistics, matrix analysis, and optimization.

Your project will produce software, educational materials, visualizations, and more. What do you hope these outputs enable researchers or students to do?

One of our goals is to expand the existing toolkit available to data analysts and quantitative researchers. We aim to develop flexible algorithms and software that can be readily used and customized by practitioners. We strive for methods that are interpretable and mathematically principled.

Talk about applications or impacts this work could potentially have across disciplines.

Together with collaborators in psychiatry and applied mathematics, we are studying the neurobiological determinants of psychosis, specifically schizophrenia. Towards this end, tools from this project will be applied to representations of brain scan images in multiple samples of people – think healthy individuals, at-risk individuals, and individuals diagnosed with schizophrenia.

In the long term, findings from this project may contribute new, principled methods that help researchers develop personalized treatments and improve mental health outcomes.

The CAREER Award program emphasizes education and workforce development. How will this award support work with students or partnerships with other institutions?

This award prioritizes the training of student researchers, both at the graduate and undergraduate level. Students in our group will participate in and help shape the overall research agenda, with opportunities to collaborate across institutions and present their findings at professional meetings. Recent alumni from our research group have gone on to graduate school and to jobs in industry, including at Meta, Uber, and LinkedIn.

What guidance would you give to students or younger researchers interested in statistics or data science?

In short, three words come to mind. Motivation. Modesty. Mentors.

---

See more recent award recipients among Statistics faculty, staff, and students.