Naiti S. Bhatt
Building reproducible AI evaluation systems using insights from developmental cognitive science and open science practices. I craft evaluation workflows, reproducible ML pipelines, and research infrastructure, and am currently focused on how insights from childhood learning and cognitive mechanisms can help us evaluate and improve AI systems.
What I work on
LLM Evaluation & Alignment
Evaluating large language models using tasks validated for benchmarking human learning and cognition across domains of social reasoning.
Research Software Engineering
Designing scalable, reproducible end-to-end analysis pipelines for processing, characterizing, and modeling human-subjects data (diffusion & functional MRI, behavioral questionnaires, eye-tracking, egocentric image comprehension).
Developmental Cognitive Science
Modeling neurocognitive processes underlying skill (object understanding, decision-making, and social reasoning) development across childhood.
A bit more about me
I completed my masters at the Department of Psychology at the University of Edinburgh, where I built reproducible analysis pipelines for functional and diffusion MRI to study the neural mechanisms of theory of mind development for my dissertation work. Before Edinburgh, I graduated from Scripps College with majors in Computer Science and Neuroscience, where I trained and tested computer vision models to segment and detect objects in infant egocentric video frames to characterize early object learning for my thesis work.
I grew up in Basking Ridge, New Jersey. Outside of research, I roast and brew specialty coffee, chase sunrises and sunsets, and swim in international waters (19 countries & 30 US states so far, from 4°C to 40°C).