Projects

Most of my projects have explored how the representation of information shapes system function by enabling certain interpretations and behaviors while constraining others. Regardless of the domain, I’ll probably always come back to this question.

LLM-based extraction and graph analysis

Just like the universe, available information keeps expanding. Yet identifying meaningful insights only becomes harder without the proper navigation tools. Crystal (2026) transforms unstructured text into a queryable knowledge graph, surfacing non-obvious relationships, patterns, and potential knowledge gaps to guide hypothesis generation across fields that may otherwise remain disconnected or overlooked by individual researchers.

Computational linguistics and machine learning

Project ESPAÑOL (2024) leverages high-dimensional analysis of classical Spanish-language poems to create custom lesson plans and guide discovery of new texts for practice and study. Using high-throughput web scraping, I generated a dataset of 10,000 Spanish-language poems in the public domain (totaling 2.5 million words), and then ran an unsupervised clustering algorithm in Python to group the poems into difficulty levels by analyzing verb frequencies and grammatical tenses. I built an interactive application to view the full texts, metadata, and grammatical statistics. As far as I know, this might be the only way to answer a question like “Which 19th-century Colombian poet used the most future-tense verbs?”

Computational publishing for creative non-fiction

Mind in the World (2025) is an essay collection exploring science, philosophy, and the experience of being human today. I wanted to write about aspects of life that don’t feel reducible to mechanistic frameworks, conveying concepts from my background in neuroscience research to a non-technical audience. I used Markdown, Obsidian, and Quarto to produce an interactive technical presentation that incorporates hyperlinks, footnotes, and references from across the Web and YouTube. I wrote most of this during a time when I was on a plane for 3+ hours twice a week, over several months.

Neural cell types and circuits for vocal learning

at University of Texas Southwestern Medical Center in the Roberts lab and Konopka lab

My post-doctoral research used comparative high-throughput transcriptomics to understand how the brain produces complex, learned behaviors like speech and language. These projects were the first to implement single-cell RNA sequencing in songbirds, and the results have broad implications for understanding the genetic toolkits that neurons and circuits use to perform advanced computations. I performed end-to-end analysis of large-scale, high-dimensional gene expression data encompassing molecular library preparation, computational pipelines, dataset integration, unsupervised clustering, and statistical modeling.

Open-access links to publications:

Neuromodulators of motivation and reward in vocal communication

at University of Wisconsin-Madison in the Riters lab

My doctoral dissertation examined the neural control of vocal communication across contexts, using songbirds as a model system. I identified neurotensin, a neuropeptide involved in motivation and reward that strongly interacts with dopamine, as a potential modulator of context-specific vocalizations. I applied multiple linear regression techniques to analyze individual, developmental, and physiological variation in gene expression, protein labeling, and behavior.

Open-access links to publications: