RAG in Azure with OpenAI and ChatGPT LLM model

About this course

Build an AI experience that can use your documents, not just the information already stored in a language model. This course takes you through retrieval-augmented generation (RAG) with Azure OpenAI and Azure AI Search, showing how document preparation, embeddings, retrieval, and a ChatGPT-style model fit into one workflow.

Follow the document-to-answer journey

Work through the individual stages in Python Jupyter notebooks so you can see what happens to your data before a question ever reaches the model. Then explore a second, interface-driven approach in Azure AI Studio. Seeing both paths helps you connect the underlying concepts with the managed tools available in Azure.

What you will work through

Understand the purpose of RAG and how retrieved content provides context for a language-model response.

Create the Azure resources used by the examples through the portal and command-line tools.

Connect to an Azure OpenAI model and make an initial request.

Count document tokens and clean Markdown files before preparing them for retrieval.

Split documents into chunks and generate embedding vectors that represent their content.

Create an Azure AI Search index and upload the prepared document chunks.

Run vector searches to retrieve relevant content for a question.

Combine retrieved document content with model prompts to build a document-grounded chat workflow.

Configure and explore the RAG experience through the Azure AI Studio interface demonstrated in the course.

Learn the moving parts of RAG

The emphasis is on understanding the pipeline, not simply pressing a button and accepting an answer. You will see where token counts, document structure, chunking, and search enter the process, giving you a foundation for investigating your own document-based AI applications.

Read the full description on Udemy on Udemy (opens in a new tab)

Prices, discounts and availability are set by Udemy. We may earn a commission when you purchase through links on this site.