Back to Portfolio

LectureNotes RAG System

Project Overview

GPU-accelerated Retrieval-Augmented Generation for Academic Documents

LectureNotes RAG System is a high-performance question-answering system designed specifically for academic documents and lecture notes. Built with GPU acceleration in mind, it leverages NVIDIA's RTX 5090 to deliver lightning-fast semantic search across thousands of documents.

The system implements a complete RAG pipeline: from multi-format document ingestion (PDF, DOCX, TXT, MD) to intelligent text chunking, high-dimensional embedding generation, and efficient vector search using FAISS. It integrates seamlessly with LM Studio for local, private LLM inference, ensuring data privacy while providing intelligent responses.

1000+
Chunks/Second
<10ms
Search Latency
1024
Dimensions
4
File Formats

Key Technologies

Python
FAISS
PyTorch
CUDA
Sentence Transformers
LangChain
Streamlit
LM Studio
BAAI/bge-large-en-v1.5

Use Cases

  • Academic Study: Query lecture notes and textbooks for quick revision
  • Research Assistant: Search through research papers and extract key insights
  • Documentation Q&A: Interactive exploration of technical documentation
  • Knowledge Management: Build and query personal knowledge bases