Projects

2026–Present • Postdoctoral Researcher, Halıcıoğlu Data Science Institute, UC San Diego

ASPIRE (Adaptive Scaffolding for Personalized Instruction and Responsive Education) studies how learners build mathematical capability with LLM-based adaptive scaffolding directed by expert knowledge graphs, across university and community college settings.

I design and analyze studies of learner progress, build and validate measures where no established instrument exists — including engagement measures derived from automated facial-expression analysis — and author research protocols for multi-site studies across UC San Diego, the San Diego Community College District, and high school pilot partners.

2025 • Researcher & Product Lead

PapyrusAI is an intelligent coach designed to meet two major challenges: improving students’ writing and AI literacy. Students log on to the PapyrusAI Web portal where they receive detailed individual feedback via GPT-4’s API on their writing plans and written texts, as well as scores per school-provided rubrics. An instructor portal allows teachers to select among existing prompt sets or add their own prompts, while also including a dashboard showing and analyzing student use.

PapyrusAI functions like a Socratic tutor, helping students develop their topic, outline, or argument through back-and-forth dialogue, and providing feedback on their organization, content, and language use. Through using it, students get first-hand practice working with AI in a safe and scaffolded environment.

2025 • Chair

AIforCA is a project of CSforCA, a coalition of K–12 educators, nonprofit organizations, higher education, and industry leaders. We believe research, data, and classroom experience should inform policy to ensure all California students have access to high-quality CS education.

We engage the community through policy discussions and teacher training events where we share lessons about AI and CS.

2023 • Middle School Lead (Former)

An integrated curriculum using an innovative Scratch-based platform to code AI chatbots, create 3D modeling, and support users with an AI assistant capable of image creation. Middle school students explore generative AI tools, build AI-powered applications, create voice and vision AI apps, and explore advanced topics like neural networks and self-driving cars.

2022 • Middle School Lead (Former)

The FairytaleQA dataset contains CSV files of 278 children's stories from Project Gutenberg and a set of questions and answers developed by educational experts based on an evidence-based framework. It focuses on narrative comprehension of K–8 students to facilitate assessment and training for both machines and young readers.

We then developed StoryBuddy, an AI-enabled system for parents to create interactive storytelling experiences, highlighting the need to balance parent involvement and bonding with minimizing intervention when parents are busy.