Nidhi Nasiar is a postdoctoral researcher with the SCALE Initiative at Stanford University. Her research includes automated detection of student behaviors, including affect, self-regulation, and disengagement, using ML- and LLM-based techniques on fine-grained K-12 data. She cares deeply about the fairness of models from design to implementation, and about how demographic categories are constructed and used for evaluating fairness. She develops and evaluates novel ways of using LLMs to automate annotation and analysis of tutoring transcripts. She has conducted dosage- and intervention-related RCTs to support student learning. Her current work focuses on developing math tutoring benchmarks, evaluating the effectiveness of different pedagogical moves, and conducting A/B tests to investigate how to develop effective AI tutors.
She is curious about what effective tutoring entails and how to provide appropriate in-the-moment support via personalized scaffolds.
She holds a PhD in Learning Sciences and Technologies from the University of Pennsylvania, where she was part of the Penn Center for Learning Analytics; an M.S.Ed. in Learning Sciences and Technologies from the University of Pennsylvania; and a B.Tech. in Electronics and Communications Engineering from LNMIIT in India.
