An AI-powered webapp designed to simplify and streamline the course scheduling process for University of Washington students.
We were motivated by the inefficiencies in UDub’s MyPlan, which makes scheduling more complicated than it needs to be. Our goal was to create a smarter, more intuitive solution that eliminates the usual scheduling headaches.
UDub-Scheduler takes the hassle out of course scheduling. With the power of Llama 2, you can simply chat with the AI about your courses, interests, and preferences. The app generates a perfect schedule instantly and even maps out course locations on a clean, interactive map—super convenient for navigating campus.
We fine-tuned Llama 2 using Intel’s AI PC, training the model with sample prompts and scheduling data. After refining the chatbot, we integrated it into a sleek, user-friendly webapp that allows for real-time schedule generation and visualization.
Getting RAG Llama to run on Intel’s cloud proved to be tricky at first. However, we adapted quickly, learning as we went, and continued making progress without skipping a beat.
We’re proud of how smoothly everything came together. By overcoming technical challenges, we’ve built an AI-driven scheduler that just works, delivering a seamless user experience.
- Multi-quarter planning
- Course recommendations based on Rate My Professor sentiment analysis
- Reducing AI hallucinations by injecting real-time course catalog and prerequisite data during inference
- Clone the repository:
git clone https://github.com/not-matty/udub-scheduler.git
