Class Intelligence System
Departmental RAG system with ChromaDB and Gemini citations.
01 // OVERVIEW
Departmental RAG system with ChromaDB and Gemini citations.
University course materials are fragmented across unsearchable slide decks, scanned PDFs, and drive folders. Students struggle to find precise syllabus answers or exam concepts before deadlines.
A high-precision Retrieval-Augmented Generation (RAG) platform tailored for academic departments, pairing chunked vector search with Google Gemini 1.5 Flash for grounded, cited answers.
02 // SYSTEM ARCHITECTURE & DATA FLOW
Document chunking and embedding pipeline indexed into ChromaDB, queried via FastAPI with cosine similarity thresholds and contextual prompt synthesis.
03 // TECHNICAL DECISIONS & TRADE-OFFS
Enables straightforward deployment without expensive managed vector database overhead for department-scale corpora.
Exceptional cost-performance ratio with low latency (sub-500ms) and large context window capability.
04 // CHALLENGES & RESOLUTIONS
Extracting clean text and equations from low-resolution faculty PDF scans.
Preventing hallucinations when questions fall outside the department syllabus scope.
05 // VERIFIED OUTCOMES
Production-ready deployment serving departmental course notes with source citations.
Custom authentication with pre-hashed SHA-256 passwords and role-based access control.