2021 · 2023
Short-answer assessment
Models that compare student answers with questions and rubrics, then group similar answers to support instructor feedback.
Related publication →Research
AI in educational practice
My work brings together natural language processing, multimodal models, and educational assessment. These projects connect technical questions to specific teaching and learning tasks.
01
Assessment & feedback
Making educational judgments useful and reviewable
I develop models and datasets for assessing short answers, STEM reasoning, and instructional materials. This work studies both prediction quality and the information a teacher needs to interpret or review an assessment.
2023 · 2024
Human review and selective prediction
Methods that identify when an automated grading system should defer to a person.
Related publication →2026
SciEval
A benchmark of 273 K–12 science lessons, assessed against 13 criteria from the EQuIP rubric.
Related publication →02
Reading & language
Tools for creating and studying reading experiences
I work with literacy researchers on story generation, dialogic reading, and evaluation of instructional materials. The focus is on how these tools fit the work of educators and the needs of beginning readers.
2025
StoryLab
A system for teachers to guide the generation of illustrated stories for young learners.
Related publication →Ongoing
AI Reading Enhancer
Ongoing work on generating decodable books and evaluating them with literacy experts.
Ask about this work →2025
Dialogic reading
Studies of bilingual English learning and comparisons between reading with a parent and a conversational agent.
Related publication →03
Multimodal learning
Understanding learning through interaction
I study the language and behavior in parent-child and child-robot interactions, combining multimodal data with questions from education and developmental research.
2025 · 2026
ASD-HI and PaiCoach
Analysis of parent-child interactions during shared book reading, with expert review of AI-supported coaching feedback.
Related publication →2026
Embodied reading companions
Feasibility research on an LLM-powered humanoid robot as a reading companion for children's AI literacy.
Related publication →2026
LLM4Art
Collaborative research on how language models interpret artworks.
Related publication →Ongoing
Dyslexia simulation
Ongoing work on interactive simulations for educator training and disability awareness.
Ask about this work →Research resources
Datasets & software
I-STUDIO
A dataset of short student answers to statistics questions, with assessment labels.
Dataset record ↗Physics lab report assessment
Data for machine learning on rubric-based assessment of physics lab reports.
Penn State Data Commons ↗EasyML
A dataflow-based system for applying machine learning algorithms to real-world tasks.
Repository on GitHub ↗