I'm a human-centered AI researcher at UC San Diego working at the intersections of human-computer interaction
and computational social science. Currently pursuing my M.S. in Computational Social Science
while conducting research at the Design Lab and Herbert Wertheim School of Public Health.
Jun 2025Graduated with B.S. in Cognitive Science from UC San Diego
Publications
CHI 2027 (In Review)
U-flex: An AI-Supported Adaptation Framework for Digital Behavior Change Interventions
Steven De La Torre, Taggert Smith, Meelim Kim, Eric Hekler
Development of an automated natural-language barrier detection pipeline and automated design adaptation framework for behavior change interventions.
@article{delatorre2027uflex,
title={U-flex: A Grounded LLM Framework for User-Driven Adaptation in Digital Behavior Change Interventions},
author={De La Torre, Steven and Smith, Taggert and Kim, Meelim and Hekler, Eric},
booktitle={Proceedings of the 2027 CHI Conference on Human Factors in Computing Systems},
year={2027},
note={In Review}
}
COLM 2026
QualAlign: Benchmarking Automated Qualitative Coding Against Human Schemas
Taggert Smith, Chuong Nguyen, Qisen Yang, Oishani Bandopadhyay, Yaru Su, Nadia Polikarpova, and Xinyu Pi
Benchmarks how well LLM-based systems perform qualitative coding against human expert schemas using lexical, semantic, and assignment-based scores, finding that LLM-generated codebooks which appear semantically similar to human work do not necessarily reflect a similar qualitative interpretation of the data at the excerpt level.
@inproceedings{smith2026qualalign,
title={QualAlign: Benchmarking Automated Qualitative Coding Against Human Schemas},
author={Smith, Taggert and Nguyen, Chuong and Yang, Qisen and Bandopadhyay, Oishani and Su, Yaru and Polikarpova, Nadia and Pi, Xinyu},
booktitle={Third Conference on Language Modeling (COLM 2026)},
year={2026}
}
ASSETS 2026
Everyday Technology Practices of Adults with ADHD: Insights for Assistive Technology Design
Taggert Smith, Andy Smithwick, Oishani Bandopadhyay, and Imani Munyaka
Using natural language processing and thematic analysis on online ADHD community discussions to uncover common frustrations, preferences, and needs relating to technology use, identifying gaps and proposing extensions to existing assistive technology design frameworks.
@inproceedings{smith2026adhd,
title={Everyday Technology Practices of Adults with ADHD: Insights for Assistive Technology Design},
author={Smith, Taggert and Smithwick, Andy and Bandopadhyay, Oishani and Munyaka, Imani},
booktitle={Proceedings of the 28th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '26)},
pages={1--18},
year={2026},
organization={ACM},
doi={10.1145/3797867.3829014}
}
Health Psychology Review
Towards an Integrative Model of Physical Activity: Bridging Idiographic and Nomothetic Perspectives Through Participatory Mapping
Meelim Kim, Olga Perski, Steven De La Torre, Taggert Smith, Junghwan Park, Daniel E. Rivera, Predrag Klasnja, Eric Hekler
Research on Just-In-Time Adaptive Interventions (JITAIs) for physical activity, integrating idiographic and nomothetic perspectives through participatory systems modeling.
@article{kim2026integrative,
title={Towards an Integrative Model of Physical Activity: Bridging Idiographic and Nomothetic Perspectives Through Participatory Mapping},
author={Kim, Meelim and Perski, Olga and De La Torre, Steven and Smith, Taggert and Park, Junghwan and Rivera, Daniel E. and Klasnja, Predrag and Hekler, Eric},
journal={Health Psychology Review},
pages={1--31},
year={2026},
doi={10.1080/17437199.2026.2738646}
}
FAccT 2025
"I Hadn't Thought About That": Creators of Human-like AI Weigh in on Ethics & Neurodivergence
Naba Rizvi, Taggert Smith, Tanvi Vidyala, Mya Bolds, Harper Strickland, Andrew Begel, Rua Williams, and Imani Munyaka
Interviewed academic and industry experts in AI, chatbots and robots, analyzing qualitative data via thematic mapping and phenomenological analysis to understand perspectives on ethics and neurodivergence in human-like AI systems.
@inproceedings{rizvi2025hadnt,
title={"I Hadn't Thought About That": Creators of Human-like AI Weigh in on Ethics & Neurodivergence},
author={Rizvi, Naba and Smith, Taggert and Vidyala, Tanvi and Bolds, Mya and Strickland, Harper and Begel, Andrew and Williams, Rua and Munyaka, Imani},
booktitle={Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency (FAccT '25)},
pages={3385--3399},
year={2025},
organization={ACM},
doi={10.1145/3715275.3732218}
}
Other Projects
Architectural Design Analysis Tool (Design Lab)
Developing an evaluation tool to predict how architectural spaces affect human psychology using VLMs for image tagging, LLMs for literature synthesis, structural causal modeling, and bayesian optimization for validation experiments.
Benchmarking Open-Source LLMs for Deductive Qualitative Coding
Replication of an LLM-assisted qualitative coding study, extended into a multi-model benchmark for closed (deductive) coding across curated thematic analysis datasets. This independent work contributes to my upcoming U-flex system and paper.
Food Assistance Dashboards (San Diego Hunger Coalition)
Creating interactive and flexible data visualization dashboards to support place-based decision-making for San Diego policymakers, hunger researchers, and community workers.
CoFurnish: Community Furniture Exchange Kiosk
A lobby kiosk and mobile-assisted flow designed to help apartment residents buy, sell, and donate furniture
during move-in and move-out periods. This project combined user research in the field, usability testing, and physical
prototyping to address the challenges of furniture exchange in apartment communities.