About Me
My interests, at their core, revolve around difficult puzzles, physical and mental alike. What I find most satisfying isn't the solution itself but the process of getting there, the kind of pursuit where effort and attention reliably translate into visible improvement. I find that feeling more motivating than any external benchmark.
Data Science & Industry
I currently work as a Principal Data Scientist II at Vail Systems in Chicago, on the Research in Artificial Intelligence in Linguistics and Systems (RAILS) team. My work there sits at the intersection of cutting-edge AI research and practical voice and language technology — specifically, exploring novel AI methods that can be applied to real-world telecommunication and speech problems.
Before moving into this role full-time, I interned at Vail as a Data Science / NLP intern, where I developed an anti-spoofing model to detect whether a voice is human or AI-generated.
Across all of these roles, my toolkit has centered on R, Python, and PyTorch, with a strong emphasis on Bayesian modeling, corpus analysis, and careful experimental design.
Research: Linguistics & Cognitive Science
My academic research lives at the boundary of psycholinguistics, cognitive science, and computational modeling. At its core, I'm interested in a deceptively simple question: how does the way we learn language shape our cognitive representations of language?
More specifically, my work focuses on error-driven learning and linguistic storage — how prediction error during language processing drives long-term changes in mental representations, and whether we store language as abstract rules, as specific memorized chunks, or as some blend of both. I approach these questions using a combination of computational modeling (e.g., neural networks), corpus methods, and experimental psycholinguistic techniques including eye-tracking.
Some of my recent work has examined how the frequency and predictability of multi-word phrases affects their cognitive representation; how frequency-dependent regularization can emerge from a noisy-channel processing model; and how large language models encode abstract vs. item-specific linguistic knowledge — and what that reveals about human language learning by comparison.
I completed my Ph.D. in Linguistics and M.A. in Psychology at UC Davis, advised by Dr. Emily Morgan and Dr. Fernanda Ferreira respectively, and I'm now a Courtesy Research Associate at the University of Oregon, where I continue to collaborate with Dr. Vsevolod Kapatsinski and the Usage-Based Linguistics Lab.
Outside the Office
🏒 I'm proudly from Connecticut, and despite living in Oregon, California, and now Chicago, there's a particular culture and identity that comes with growing up in New England, one that stays with you wherever you end up. That includes a lifelong, often painful allegiance to the Boston Bruins.
🤺 I fenced competitively for most of my life — foil, if it matters, which it does — eventually captaining the UC Davis men's foil team and later coaching at Davis Fencing Academy, and still fence recreationally today.
🎹 I've also played piano for most of my life, and it remains one of the few pursuits that can humble me on any given day.