PhD Student · MIT Media Lab
I'm a PhD student at the MIT Media Lab, advised by Paul Pu Liang in the Multisensory Intelligence group and co-advised by Pattie Maes. I work at the intersection of multimodal AI and human-computer interaction. My research focuses on breaking health-related socio-cultural misinformation by building interactive multimodal systems like video agents that adapt across demographics to make medical information culturally resonant for the communities they reach.
I spent the summer of 2026 training LLMs at IBM Research with Rogerio Feris, and continue as a visiting student at the MIT–IBM Watson AI Lab. Before that I spent a summer at Adobe Research working on multimodal mathematical reasoning, and was a research specialist at the AI Institute, University of South Carolina, with Amit Sheth.
A field study with 434 women in suburban India. Presenters that matched a community's culture cut false health beliefs by 30% — roughly double a culturally neutral presenter — and the effect held three weeks later.
A benchmark for whether video generation models depict a culture faithfully or fall back on stereotype.
Over four weeks, 67 participants judged news headlines and images. Chatbot help made them 21% more accurate in the moment, but built no discernment that survived once the help was taken away.
An attribution mechanism that makes multimodal models show which parts of a chart actually support their answer, across visual question answering, reasoning, and attribution.
Reframes deception around lies of omission — their types, colors, intentions and topics — with an NLP model built to catch them.
Fact verification decomposed into Who, What, When, Where and Why, so a system can point to exactly where a claim and its evidence diverge.
County-level emotional response to the US COVID-19 vaccine rollout, and whether anger, anxiety and incivility in vaccine tweets predict local hesitancy and vaccination rates.
A benchmark and detectability index for AI-generated text in Hindi, ranking LLMs by how detectable their Hindi output is.