QUD-Empowered LLM Essay Co-writing Tool

Built a comprehensive essay co-writing tool that uses the linguistic Questions Under Discussion (QUD) framework to generate pedagogically appropriate questions that scaffold student learning. Developed a FastAPI backend to create golden QUD questions and run coverage analysis of student summaries using OpenAI GPT-4o, embedding similarity, and LLM-as-a-judge verification. Created a React/Vite/Tailwind study platform with access control, progressive article chunking, and randomized question sets; deployed on Cloudflare Pages with API on Railway/Render. Completed Study 1 question ranking with significant correlation between AI rankings and human importance ratings (r = 0.28, p = 0.002), and designed Study 2 for human preference evaluation. Masters Capstone project with team members Neng Wan, Kiara LaRocca, and Ting-Yu Chou.