C@B_GPT

An AI-powered guide to course registration at Brown—analyzing your academic history, requirements, and schedule to craft the smartest path forward, that is tailored towards your graduation goals.

Comment

Inspiration

Our inspiration stems from our own experience with the challenges of course registration at Brown University. The overwhelming number of course options often leaves students struggling to make confident decisions, turning registration into one of the most stressful periods of the semester. We believe that academic planning should be a long-term, strategic process over four years rather than a rushed decision each semester. By bridging the gap between students and the university’s course offerings, we aim to create a more streamlined, informed, and personalized registration experience—empowering students to make better academic and career-aligned choices.

What it does

C@B_GPT is an LLM-powered course registration assistant that provides students with personalized guidance based on their academic history, degree requirements, and scheduling constraints. It enables students to: • Query degree requirements dynamically and receive tailored recommendations based on their completed coursework. • Explore interdisciplinary courses that align with their interests and career aspirations, leveraging Brown’s open curriculum. • Receive intelligent course suggestions, helping students optimize their schedules and find classes that complement their academic goals. For example, a student interested in computer science and neuroscience can use C@B_GPT to discover courses at the intersection of these disciplines, ensuring a well-rounded academic experience.

How we built it

We utilized a modern AI and web development stack to build C@B_GPT, integrating:

  1. OpenAI API (GPT-4o Mini) – Powers the natural language processing for course-related queries.
  2. LangChain – Facilitates seamless retrieval-augmented generation (RAG) and API interaction.
  3. Flask – Provides a lightweight Python-based backend for serving API requests.
  4. Web Scraping techniques – Extracts course registration data from Brown’s central registration system, professor pages, and degree requirement documents for enhanced contextual accuracy.

Challenges we ran into

We encountered several technical hurdles, particularly in implementing Retrieval-Augmented Generation (RAG) effectively: • Output formatting issues – The LLM occasionally structured responses incorrectly, requiring fine-tuning. • Scalability of data ingestion – Handling large amounts of registration data efficiently posed infrastructure and optimization challenges. Each challenge pushed us delve deep into the LLM complexities and develop efficient fallback mechanisms to improve response accuracy.

Accomplishments that we're proud of

We believe C@B_GPT is a milestone in education technology, particularly for Brown University, due to:

  1. Solving a real pain point – Course registration is a widespread challenge, yet students have traditionally relied solely on the school’s central portal with limited personalized guidance. Our tool pioneers a new approach by introducing an intelligent, student-centric resource that enhances decision-making—something that has been largely overlooked until now.
  2. Tailored Personalization – Every recommendation given is personalized to align with each student’s academic and personal interests.
  3. Scalability & Integration Potential – The system is designed to be expandable beyond Brown, with potential integration into cab.brown.edu or other institutions’ portals.

What we learned

Working with LLMs reinforced that while they are powerful for natural language understanding, they suffer generalizability. However, by leveraging Retrieval-Augmented Generation (RAG), we are able to leverage its advantages as well as supplement on its weaknesses to tailor it to perform customized tasks, and in this case, to help students with course registration.

What's next for C@B_GPT

Currently, C@B_GPT primarily caters to STEM students at Brown. Our next steps include: • Expanding to all academic disciplines, ensuring comprehensive coverage for humanities, social sciences, and arts. • Extending to other universities, creating a generalized AI-powered academic advising system applicable across different institutions.

Built With

Share this project:

Updates