Our project focuses on building a Retrieval-Augmented Generation (RAG) system that allows users to search through a collection of DS4300 course notes and receive AI-generated responses based on relevant material. The system ingests and indexes documents, retrieves context from a vector database, and uses a locally running LLM to generate responses.
We are experimenting with different chunking strategies, embedding models, vector databases, and local LLMs to compare their performance in terms of retrieval accuracy, speed, and memory usage. The goal is to find the most effective pipeline for organizing and accessing course material, making it easier to find relevant information.