Faculty-sponsored RAG system
Northeastern University Assistant v2.0
Retrieval over 80,000+ university pages, answering student questions with a cited source.
- Pages indexed
- 80,000+
- 2025
The problem
Students could not get straight answers about programs and policy; the information was spread across hundreds of university pages.
What made it hard
An answer without a citation is worse than no answer, so every response has to name the page it came from.
Architecture
- 01Scrapy crawls and normalises university pages into a corpus
- 02ChromaDB stores embeddings; queries are expanded into variations
- 03Hybrid retrieval combines semantic and keyword search
- 04FastAPI serves answers with a confidence score and source link
Outcomes
- Cited by Northeastern's 2026 Outstanding Master's Student Award for Community Impact
- Faculty-sponsored and deployed for student use
- 80,000+ pages indexed and searchable