1st Runner Up — AI Thailand Hackathon 2024 EP2

Competed as Team Outlier in the AI Thailand Hackathon 2024 — AI for Thai API on Shelf, held at Sirindhorn Science Home. Out of 28 teams in the final round, our team placed 1st Runner Up with a 36-hour build.
We developed an application that automates exam creation from uploaded course materials. Educators upload a PDF — lecture notes, textbooks, or syllabi — and the system generates structured exam questions in minutes, a process that traditionally takes weeks of manual work. The application leveraged the AI for Thai API for Thai-language NLP processing.
Beyond generation, the system implemented an Elo-based adaptive difficulty mechanism inspired by competitive chess ratings. As students answer questions, their proficiency rating adjusts dynamically — the system selects harder or easier questions to match their level, keeping assessments in the zone of productive challenge. Questions were also classified across Bloom's Taxonomy levels (Remember, Understand, Apply, Analyze, Evaluate, Create), allowing educators to ensure their exams tested higher-order thinking rather than just recall.
This project became an early exploration into AI-assisted education workflows, and the ideas around automated assessment directly influenced the later development of the CULI platform.





