Explore the interactive country-aware atlas of NLP dataset representation with searchable dataset exploration, geographic analyses, visualizations, and downloadable metadata covering 13,000+ dataset records across 195 countries. Check it out at AtlasNLP!
AtlasNLP: A Country-Aware Atlas of Dataset Representation in NLP builds on my digital divide research with a large-scale pipeline that audits dataset representation across nearly 120,000 ACL papers, identifying and characterizing 13,000+ paper-level dataset records and analysis gaps. Really excited to share this work!
I look forward to contributing to interdisciplinary work on participatory AI, unresolved NLP research problems, fairness, and poverty-focused AI.
This summer in Baltimore, I designed a scalable alternative to one-off traffic studies by reusing footage from Boston's approximately 700 camera-equipped intersections in collaboration with the City of Boston's Department of Innovation. I learned so much from the program and the amazing cohort!
I spoke on "Applying for Ph.D. Programs in Machine Learning/AI". I really enjoyed sharing my experiences and insights with aspiring graduate students in the community.
This honor from the University of Michigan recognizes outstanding academic achievement and means so much to me.
I will be working at Johns Hopkins University this summer on designing scalable solutions for traffic studies in collaboration with the City of Boston's Department of Innovation. Can't wait to get started!
Dive into the interactive public dataset resource for culturally grounded, function-centric visual representation across over 60 cultures. This is a culmination of our work on cultural diversity in AI! Explore it at Culture Affordance Atlas!
At the NeurIPS 2025 Workshop on Socially Responsible and Trustworthy Foundation Models .
At the African Women Film Festival 2025 on the panel for the Technologies, Artificial Intelligence, and the future of Filmmaking in Africa discussion.
Culture Affordance Atlas: Reconciling Object Diversity Through Functional Mapping will be presented at AAAI 2026. See you at Changi, Singapore! This is work is a nice follow up on the 'Bridging the Digital Divide paper and Uplifting Lower Income Data'. Check it out!
Uplifting Lower-Income Data: Strategies for Socioeconomic Perspective Shifts in Large Multi-modal Models will be presented at NAACL 2025. See you at Albuquerque, New Mexico! This is work is a nice follow up on the 'Bridging the Digital Divide paper'. Check it out!
I am learning a lot from the students and my fellow instructors in the Information Retrieval course.
Why AI Is WEIRD and Should Not Be This Way: Towards AI For Everyone, With Everyone, By Everyone
CVQA: Culturally-diverse Multilingual Visual Question Answering Benchmark
Has It All Been Solved? Open NLP Research Questions Not Solved by Large Language Models
Annotations on a Budget: Leveraging Geo-Data Similarity to Balance Model Performance and Annotation Cost
I am now a PhD Candidate!
Bridging the Digital Divide: Performance Variation across Socio-Economic Factors in Vison-Language Models will appear at the 2023 Conference on Empirical Methods in Natural Language Processing. See you in Sentosa Island, Singapore!!