I am investigating what information people want AI agents to retain across data workflows, how they want to manage it, and how it relates to their goals and values. I will develop and evaluate an agent harness that enables users to inspect and control the context agents use across data-related tasks.

Conducted a user study examining how novice and expert users interact with LLMs when debugging NL2SQL queries, with the goal of informing more effective human-AI collaboration in data-driven tasks.

Exploring the human-AI interaction layer of a explainable multimodal DBMS. Interested in how users query, debug, and interpret multimodal results, and how the system can best surface explanations and lineage to support understanding and trust.

Explored how virtual fidgets could support attention during online lectures. Through a study with eight students, we identified eight themes in their experiences and proposed four design principles for future virtual fidgets.

Created an interactive dataset visualization tool with R to enable environmental justice community groups to explore air quality concerns in their community with the Project Eclipse sensor data.

Aimed to reduce impostor syndrome in underrepresented students through strengthening data structure skills with an agumented reality application.

Assisted with the qualitative coding of an exploratory survey and participatory design workshops aimed at gathering insights of Black and Latinx students' perspectives of campus services accessibility.

Attempted to provide probabilistic forescasts of sea levels using Gaussian Process, Neural Networks, and different variations of these models. (Github Repo)