I’m a master’s student in Computer Science at Stanford University, advised by Yejin Choi and Douwe Kiela. Currently, I’m doing alignment research with Dylan Hadfield-Menell at MIT.
I’m interested in how language models can adapt to different people, values, and preferences. My research interests include pluralistic alignment, personalization, multi-agent systems, post-training (RL), synthetic data, LLM evaluation, and representation learning.
Previously, I worked with the Alexa and AWS AI teams.
Pluralistic alignment
Pluralistic Alignment @ ICML · 2026
SCPO trains reward models for diverse cultural preferences by filtering and weighting country-specific feedback, improving data efficiency while moderating excessive preference bias.
Persona retrieval
EMNLP (Oral) · 2023
PK-ICR jointly retrieves persona and knowledge for grounded dialogue by adapting neural question-answering models, and introduces a ranking test for hard-negative examples.
Dialogue representations
ACL (Oral) · 2026
TaDSE uses dialogue templates and synthetic utterance augmentation in contrastive learning to improve sentence representations, evaluated across five benchmark datasets for dialogue tasks.