Navigating a massive university ecosystem can be overwhelming. Information about schedules, event locations, and academic resources is often scattered across different portals and PDFs. We were inspired to create a "central nervous system" for the campus—a single, intelligent interface where students can get instant, accurate answers to any campus-related question.
CU UP campus brain acts as an AI-powered knowledge hub. It allows students to: Ask natural language questions about campus navigation and building locations. Get real-time updates on club events and workshops. Query academic handbooks and syllabus details without digging through folders. Summarize long university announcements into digestible bullet points
The core "brain" was built using a RAG (Retrieval-Augmented Generation) architecture. Frontend: Built with React and Tailwind CSS for a responsive user interface. Backend: Powered by FastAPI to handle requests. AI Logic: The Gemini API (or OpenAI) was used for natural language processing, along with Pinecone as a vector database to store campus documentation. Data Processing: Python scripts were used to parse and clean unstructured data from university PDFs and websites.
One of the biggest hurdles was data ingestion. University documents come in various formats, often with complex tables. Different "chunking" strategies were tested to ensure the AI didn't lose context. Latency issues when querying the vector database were solved by optimizing the indexing strategy.
The accuracy of the "Brain" is a key accomplishment. It can correctly interpret vague student queries and provide recommendations. The user interface was kept clean and intuitive.
This project highlighted the importance of data quality. An LLM is only as good as the context provided. Skills in vector databases were improved, and how to build a scalable API that can handle multiple concurrent student queries was learned.
The vision is to integrate CU UP campus brain into the university's official app. Future plans include: Voice Integration: Allowing students to "talk" to the campus brain. Personalized Schedules: Connecting to student IDs for personalized reminders. Multilingual Support: Helping international students navigate the campus.
Built With
- css
- flask
- framework
- html
- javascript
- python
Log in or sign up for Devpost to join the conversation.