How does anyone know anything? We collect experiences and moments, little impressions of the world that inform our thoughts and future actions. These small artifacts and lessons we accumulate throughout our lives coalesce to form the lens through which we see our reality. The words we learn as children, the language in which we choose to express ourselves, shape our view of the world around us. Every second billions of signals are sent by neurons through our brain, enabling our sense of reality, our ability to communicate, and our ability to think.
How do we know our perception is true? Though our ability to learn from the world is powerful, we are often blinded by the sheer randomness of experience. We need a filter—a way to discern whether we’ve found a genuine discovery or if we're simply being tricked by an illusion. Statistics provides a path to that clarity. By studying the chaos around us, it gives structure to our perception and allows us to quantify our uncertainty.
How do we discover something new?
We experiment and observe the world around us, recording data along the way. But data is ugly. Data is frustrating. Data by itself doesn’t reveal or illuminate anything. To make knowledge accessible, we must do more than just collect it; we must curate it. We must transform complex sets of data into visual, compelling narratives that allow others to see the patterns we have uncovered.
This portfolio tells a story of data. Data that is used to learn, teach, and persuade. Through my research in computational neuroscience, I extract representations of the world from large language models and use them to map the brain’s operations. My work centers on interpretability—designing experiments around a more accessible, dissectable model of language processing to better understand the human brain. I’m drawn to work that facilitates a better understanding of the brain because that’s what makes us human. The very basic elements of our lives (our ability to talk, our ability to think, our ability to feel) that we so easily take for granted are still poorly understood and I’m fascinated by the process of unraveling them—bit by bit.
The quest for greater interpretability doesn’t stop at experiment design; it requires taking these results, which are often complicated and unintuitive, and distilling them into accessible formats, most commonly data visualizations. As a science communicator, I believe that data visualization is one of the most compelling ways to build stories that are accessible to all, regardless of their background. For technical audiences, effective data visualizations allow for a clear overview of the work and its results. For more general audiences, it allows me to guide them through the project and highlight exciting results without any additional background knowledge.
The best way to learn is through doing. This portfolio seeks to empower you to learn about my work in a variety of ways. You can either browse for a specific artifact that catches your eye or explore further dimensions of the data by clicking below.