Working at the intersection of human-computer interaction, natural language, and accessibility — designing interfaces grounded in real user research.
A mobile app that flips the traditional ASL dictionary: learners look up unfamiliar signs by handshape, movement, and orientation instead of guessing the English word first. Built through user research with Deaf native signers, ASL learners, and interpreters.
Read case studyPrototyping a computer vision tool that detects signing activity in egocentric eye-tracking footage, automating annotation workflows for longform child language datasets. Building end-to-end pipelines for multimodal audio, video, and eye-tracking data used in sign language acquisition research, and partnering with Bay Area Deaf community organizations on accessibility-centered participant outreach.
Contributed to a video-based AI triage system built with clinicians and computer vision researchers in Stanford Hospital's emergency department. Interviewed and filmed hundreds of patients to build the foundational training dataset, then analyzed 1,000+ patient videos through symptom tracking, image and metadata annotation, and database organization in Box, REDCap, and Epic. Authored SOPs and onboarded new research assistants to scale the team's data collection capacity.
Watch project videoCoached two cohorts of middle and high school students through the pilot of the NSF-funded Ujima GIRL/GIRL+ camps — a $2.4M program designed to expand inclusion and equity for Black and African American girls in STEM. Taught programming in RoboBlocky and Ch/C/C++, sensor-based robotics with Arduino, and film production, and guided student teams in producing short films for the RoboPlay Video Competition. Facilitated visits to university research labs and activities focused on teamwork, leadership, and breaking gender stereotypes in STEM.
An agent that recommends films and books tickets through natural dialogue. Item-item collaborative filtering generates personalized picks from rating data, while a DSPy architecture handles conversational memory, multi-step planning, and tool calls to scheduling, pricing, and booking APIs.
Read case studyAn immersive mixed-reality prototype built at Immerse the Bay, Stanford's XR hackathon. A weekend build exploring spatial interaction and presence — placing memories as objects you can walk around, reach for, and arrange in physical space.
Prototype · demo on request