Building Agentic RAG from scratch - A Youtube playlist

In this edition we will talk around my Youtube playlist on "Building an Agentic Retrieval Augmented Generation framework" from scratch. The article will cover each episode step by step below.

Episode 1: Introduction

Introducing the Playlist and goals

Welcome to the first episode of the “Building an Agentic Retrieval Augmented Generation framework” playlist. We will be introducing the playlist and setting up the expectations around what we are planning to build in this entire playlist.

We will be working with the LangGraph framework to build a multi-agent workflow to create an RAG.

Episode 2: Discussing the concept of Nodes, Edges and Shared State in LangGraph

In the second episode of this playlist, we will be discussing the concept of Nodes and Edges in LangGraph and how you can build a multi-agent workflow through them.

We will also discuss the concept of shared State in LangGraph.

Episode 3: In-memory Vector Store and building an External Knowledge base

In the third episode of this playlist, we will be building an in-memory vector store that can store our external knowledge base.?

Further we will retrieve the relevant documents from the Vector store on the basis of the question asked.

Episode 4: Retrieving documents from the Vector Store

In the fourth episode of this playlist we will see our Vector Store in action. We will execute the node built previously and see what the output of the vector store looks like.

Episode 5: Creating an LLM Model

In the fifth episode of this playlist we will create a node that initializes an Open AI LLM model which will be further used by multiple nodes in the graph.

Episode 6: Building a Document grader and Grading the retrieved document from Vector store

In the sixth episode of this playlist, we will create a node that gardes the document retrieved from the vector store. The node will return a binary output of yes or no depending on whether the document was relevant to the question or not.

Episode 7: Concept of Conditional Edges in LangGraph

In the seventh episode of this playlist, we will build a conditional edge that decides whether to go ahead and generate answers or not. The decision is made on the basis of the presence of relevant documents.

Episode 8: RAG framework that generates Answers on Human prompts

In the eighth episode of this playlist, we will create a RAG framework that generates answers on the basis of Human questions by taking an external knowledge base as a context.

Episode 9: Evaluating the RAG framework performance and Grading the generated answer

In the ninth episode of this playlist, we will create a node that grades the answer generated by the RAG framework. The node will generate a binary output of yes or no, depending upon whether the generated answer was relevant to the human question or not.

Episode 10: Setting-up the Graph and Compiling the workflow in LangGraph

In the tenth episode of this playlist, we will put all the nodes together and connect them via edges to compile our multi-agent workflow in LangGraph.

Episode 11: LangGraph and Retrieval Augmented Generation framework in action

In the eleventh episode of this playlist, we will see our workflow in execution. We will see how our RAG framework performs against different Human Prompts and evaluate the performance of the workflow.

Episode 12: Visualizing the compiled Graph through Mermaid

In the twelfth episode of this playlist, we will see how we can visualize our entire workflow graph through a library called Mermaid.

Conclusion

You can check my Youtube channel for more relatable technical content.

Meanwhile what you all can also do is to Like and Share this edition and my channel among your peers and also Subscribe to my channel so that you all can get notified when I come up with more content in future.

Share and Subscribe to this Newsletter with anyone who might be benefitted from this content.

Until next time, Dive Deep and Keep Learning!