I am a Ph.D. candidate in Urban Science at MIT, advised by Professor Andres Sevtsuk. Working across urban studies and computational social science, I develop and critically evaluate AI methods for understanding human behavior and social life in cities. My central question is how emerging computational methods can produce credible and actionable knowledge about urban life.
My dissertation, Seeing HAI, connects street-level imagery, observations of human activities and interactions in public space, and GIS analysis to study sidewalk life across cities. MINGLE, my method for detecting socially interacting groups, laid part of its methodological foundation. I also investigate the reliability of computational evidence: Clarity or Confusion examines what computer vision measures contribute to urban research, while Sidewalk Moments tests whether richer visual and multimodal representations better reflect human judgments.
I spent three summers as an Applied Scientist Intern at Amazon, working on multimodal learning and reasoning evaluation at AGI and cross-modal video retrieval at AWS. As a visiting researcher at UCLA, I contributed to the 2D semantic component of WalkOCC, a sidewalk-robot perception project. Before returning to MIT, I founded CitoryTech, an urban analytics company, and led its research and consulting work for nearly six years. Our projects served municipal and private-sector clients, spanning spatial analysis, pedestrian environments, and infrastructure assessment. These included road and sidewalk asset assessment in Ulaanbaatar under a World Bank-supported infrastructure program.
At MIT’s City Form Lab, I contribute to pedestrian-flow research and lead the MIT side of Sidewalk Ballet, a collaboration exhibited at the 2025 Venice Architecture Biennale. My publications span urban research and AI venues, including Nature Cities, Cities, AAAI, and ECCV. I have also supported courses in urban design, GIS, spatial analysis, and comparative land use and transportation at MIT.