Mastering Retrieval-Augmented Generation (RAG)

Name: Mastering Retrieval-Augmented Generation (RAG)
Rating: 4.3 (109 reviews)

Master RAG from Zero to Hero: Build Real-World AI with Retrieval-Augmented Generation

Created byLuka Anicin

Last updated 10/2025

English

What you'll learn

Core principles of Retrieval-Augmented Generation (RAG) – Understand how RAG combines retrieval and generation for improved AI responses.
Implementing basic and advanced RAG architectures – Step-by-step guides to setting up RAG, Multi-Query RAG, RAG Fusion, and HyDE RAG.
Working with OpenAI embeddings and Pinecone – Practical exercises in connecting embeddings with vector databases for efficient retrieval.
Multi-query and RAG Fusion techniques – Learn strategies for better, contextually accurate answers through fusion and multi-query models.
Building and deploying RAG with FastAPI on Google Cloud Platform (GCP) – End-to-end deployment guidance for scalable RAG applications.
Prompt routing and database management – Gain experience with routing strategies and optimized content indexing for more efficient RAG systems.
Prompt caching and optimization techniques – Discover ways to reduce costs and improve response speed with caching in RAG models.

Course content

6 sections • 32 lectures • 2h 31m total length

Welcome to the RAG Masterclass course!1:52
Learn to build and deploy retrieval augmented generation solutions from zero to cloud, using OpenAI's API and Pinecone, then scale and deploy on Google Cloud Platform.
How to follow the course?3:59
Start with module two to grasp embeddings and their similarity, then move through modules three to five on advanced rag techniques, flow routing, and deployment to GCP.
Where to find materials?0:11

What is Retrieval Augmented Generation (RAG)?6:30
Learn how retrieval augmented generation (rag) works and when to use it, building from scratch with OpenAI's API and pinecone, using embeddings to retrieve relevant documents.
What are Text Embeddings and how to use them?6:53
Learn what text embeddings are, how they convert words, sentences, and documents into vectors, and how cosine similarity and vector databases enable retrieval to answer user queries with an LLM.
Building an old, deterministic F&Q chatbot4:17
Explore building a deterministic faq chatbot using a predefined faq database, choosing between a clickable faq page or a chat bubble, and integrating OpenAI embedding API to generate vectors.
How to use OpenAI's Embedding API?6:58
Learn to build scalable retrieval augmented generation with OpenAI's embedding api by installing the OpenAI library, creating a client, sending embeddings requests, and handling 1536-dimension vectors.
[EXERCISE] How to find similar text by using Text Similarity - part 14:52
Learn to use text embeddings and cosine similarity to measure document similarity, and implement an OpenAI embedding function to build a vector database for finding similar text.
[SOLUTION] How to find similar text by using Cosine Similarity - part 27:14
Learn to compute cosine similarity with numpy for embedding comparisons, build a vector database, and retrieve the most similar faq to answer questions with a scalable e-commerce chat bot.
Building a simple Chatbot using OpenAI's API7:16
Build a simple retrieval-augmented generation chatbot with OpenAI's API, embedding, and an explicit system prompt to answer ecommerce policies, including 30-day returns for ebooks and IT courses.
[EXERCISE] Building our first RAG-Based Chatbot! - part 12:39
Construct your first retrieval-augmented generation chatbot using an embedding model, cosine similarity, and a faq context to augment the system prompt and generate accurate answers.
[SOLUTION] Building our first RAG-Based Chatbot! - part 25:14
What are Vector Databases and where to store our vectors?4:46
Introduction to Pinecone (Vector Database)4:16
Putting everything together - Building a RAG with external data base - part 15:46
Putting everything together - Building a RAG with external data base - part 25:19

[EXERCISE] What is Multi-Query RAG?5:17
[SOLUTION] Building Multi-Query RAG from scratch4:51
[EXERCISE] What is Fusion RAG?2:34
[SOLUTION] Building Fusion RAG from scratch8:14
Build a fusion retrieval-augmented generation (rag) system from scratch by generating candidates, retrieving top documents, and applying a reciprocal fusion ranking to produce focused context-driven answers.
[EXERCISE] What is HyDE RAG?3:25
Explore HyDE RAG, which generates a hypothetical answer to enrich context before embedding, improving speed and relevance compared to other RAG methods.
[SOLUTION] Building HyDE RAG from scratch3:56
Build HyDE RAG from scratch by generating a hypothetical answer, embedding candidates, and retrieving the best-matching context to create an augmented prompt for deployment to GCP.

[EXERCISE] What is Prompt Flow Routing RAG?6:35
Learn prompt flow routing in retrieval-augmented generation to improve ranks by classifying user intent (factual, explanation, guidance) and routing to the appropriate prompt or vector databases.
[SOLUTION] Implementing Prompt Flow Routing RAG from scratch5:17
Implement prompt flow routing for retrieval augmented generation from scratch; classify intent into factual or guidance, use system prompts and few-shot prompts, generate json, and explore database routing in videos.
[EXERCISE] What is Database Flow Routing RAG?4:26
Route a user query to the right vector database in a rag setup. Create and upsert pinecone indices for finance, tech, and product, then generate responses with a language model.
[SOLUTION - part1] Implementing Database Flow Routing RAG2:20
Implement a routing rag by creating three Pinecone indexes and classifying inputs into product, finance, tech, or other to retrieve top three matches.
[SOLUTION - part2] Implementing Database Flow Routing RAG3:06
Compose a simple prompt to classify queries, run a vector query embedding with top 3 results from get index, and use advanced database routing RAG with NS1 namespace and metadata.

RAG Deployment Code Walkthrough8:51
Walk through deploying rags with a simple FastAPI template, Dockerization, caching, and environment configs, then run the app on GCP Cloud Run using a RAG model pipeline.
What is Prompt Caching?4:29
Learn prompt caching for rag applications, using an in-memory cache to serve repeated prompts and cut API costs, while recognizing exact-match limits and pinecone as an alternative.
Testing the RAG Deployment locally - Using Docker3:21
Test the retrieval augmented generation system locally using docker; build and run the image, verify via postman, and confirm caching behavior before deploying to cloud run.
Deploying RAG to GCP Cloud Run6:30
Deploy your locally working rag service to the cloud using GitHub Actions and Google Cloud Run, enabling the Cloud Run API and continuous deployment.

Requirements

Basic knowledge of Python programming
Prompt Engineering: Writing basic to intermediate prompts
Understanding of machine learning fundamentals

Description

Welcome to "Mastering Retrieval-Augmented Generation (RAG): From Zero to Hero"!

This course is your all-in-one guide to understanding and implementing Retrieval-Augmented Generation (RAG) — a game-changing approach to enhance AI responses with powerful retrieval capabilities. Through hands-on projects, real-world exercises, and step-by-step tutorials, you'll quickly learn how to leverage RAG architectures to build effective and scalable AI solutions.

This course is designed for AI practitioners, data scientists, machine learning engineers, and developers with a background in Python programming and a basic understanding of machine learning and NLP concepts.

What You'll Learn:

- Core RAG Architecture – Understand how RAG works, from basic concepts to advanced multi-query, Fusion, and HyDE architectures.

- OpenAI Embeddings and Pinecone Integration – Learn how to connect OpenAI embeddings with Pinecone for efficient content retrieval.

- Building RAG Models from Scratch – Implement multi-query and Fusion RAG models with hands-on exercises.

- Advanced RAG Techniques – Explore database and prompt routing, caching, and deployment for optimized RAG solutions.

- Deploying on Google Cloud Platform (GCP) with FastAPI – Deploy your RAG models in a scalable cloud environment with detailed deployment instructions.

Who This Course is For:

This course is ideal for those with a background in software engineering, Python programming, and basic ML knowledge who are eager to dive into RAG applications. It’s packed with exercises to build your expertise from scratch, making it suitable for those new to RAG while being comprehensive enough for seasoned AI practitioners looking to expand their skills.

Join us and become proficient in RAG, from setting up basic architectures to deploying scalable, real-world AI solutions!

Who this course is for:

AI practitioners who want to deepen their expertise in Retrieval-Augmented Generation (RAG) and apply it to enhance AI-driven solutions.
Machine learning engineers looking to implement advanced RAG techniques like multi-query and RAG Fusion to improve model performance.
Software engineers seeking to expand their skills by building and deploying RAG models using tools like Pinecone and FastAPI.
Data scientists interested in integrating RAG architectures into data-heavy applications for more effective information retrieval and generation.
Developers working with OpenAI embeddings and vector databases who want hands-on practice connecting these tools within a RAG framework.
Professionals aiming to deploy machine learning models on Google Cloud Platform (GCP) with an emphasis on scalability and efficient architecture.
Learners eager to master indexing, prompt routing, and caching techniques to create optimized, cost-effective RAG-powered applications.

Mastering Retrieval-Augmented Generation (RAG)

What you'll learn

Explore related topics

Course content

Introduction3 lectures • 6min

Introduction to RAGs13 lectures • 1hr 12min

Advanced RAGs: User query manipulation6 lectures • 28min

Advanced RAGs: Flow Routing5 lectures • 22min

Deploying RAGs to the cloud4 lectures • 23min

Bonus1 lecture • 1min

Requirements

Description

Who this course is for: