Boost RAG with Chroma

Overview

Boost RAG with Chroma is an intermediate, hands-on course designed for developers and AI practitioners who need to solve one of the biggest challenges with Large Language Models: their tendency to hallucinate. This course moves beyond theory and teaches you how to build a practical, effective Retrieval-Augmented Generation (RAG) pipeline to make your LLMs more trustworthy and enterprise-ready. You will learn the architectural patterns for using a vector database to create an external knowledge base that grounds an LLM's responses in verifiable data. Using a project-based approach, you will implement this pattern, drawing on the popular open-source tools Chroma and LangChain as concrete examples. The course culminates in a hands-on evaluation where you will directly compare your model's answers—with and without RAG—to qualitatively measure the improvement in factuality. You'll leave with a portfolio-ready project and the ability to build safer, more reliable generative AI applications using any set of comparable tools.

Syllabus

Building the RAG Pipeline

In this module, you will build the core of a modern, factual AI system. You'll first learn why even powerful LLMs fail and how the RAG architecture solves this by grounding them in real data. Then, you'll get hands-on experience implementing this pattern to construct a functional pipeline that connects a custom knowledge base directly to an LLM.

Evaluating Hallucination Reduction

With a functional pipeline built, this module focuses on the most critical step: proving its value. You will learn how to systematically test your RAG system, compare its grounded answers against a baseline LLM's responses, and qualitatively evaluate the reduction in hallucinations. This module culminates in a final project where you will formally document your findings.