

Tools Used: React/TypeScript, OpenAI API, Python/Flask, MongoDB, Redis, Docker, Figma, Git, Duke OKD
Our project integrated Duke's course catalog with a fine-tuned version of OpenAI's gpt-3.5-turbo model. We semantically searched our database for vector representations closest to the user's query (RAG) to generate relevant courses. Throughout the summer, we worked closely with the Duke University Registrar and the Academic Advising Center to make a tool that was accessible, useful, and accurate for students and advisors alike.