Module 1: Introduction to generative AI 

• Applications and impact of Generative AI
• Evolution and Architecture of Generative AI
• How does LLM work? 
• Different types of Generative AI 

• Text generation 
• Image generation 
• Audio and speech recognition 
• Multi-modality 

Module 1.1: Foundation Models and Transformers 

• What are Foundation Models? 
• Transformer architecture (attention mechanism, encoder-decoder) 
• Pre training vs Fine tuning 
• Open-source vs closed-source LLMs (LLaMA, Mistral, Claude, GPT, etc.) 
• Hans-on: Load and infer with Hugging Face models (e.g., BERT, GPT-2) 

Module 2: Prompt Engineering, RAG and Lang chain 

• • Introduction to Prompt Engineering
  • How Tokenization work? 
  • Necessity of Prompt Engineering 
  • Basic Prompt Structure 
  • Prompt design (zero-shot, few-shot, chain of thought) 
  • Clear and direct instructions 
  • Assigning Roles (Role Prompting) 
  • Splitting Data from Instructions 
  • Formatting Output using prompt 
  • Step by step using Precognition 
  • Using Examples in prompt 
  • How to Avoid Hallucinations 

• Project: Building Complex Prompts (Industry Use Cases) 

• Project: Compare outputs from different LLMs on same prompt 

• Role of system prompts 
• RAG architecture 
• Vector databases: FAISS, Chroma, Weaviate 
• Lang Chain, LlamaIndex overview 
• Embedding vs Image generation vs Text and Code generation
• How to improve LLM results? 

• Prompt Engineering with Context 
• Retrieval Augmented Generation (RAG) 
• Fine-tuned model 
• Trained model 

• How does it relate to RAG, GraphRAG, KAG, and CAG 
• Introduction to Lang Chain 
• Choose the best foundation model for your needs 
• Explain ability and Interpretability 
• Hands-on: Build a RAG chatbot using OpenAI + FAISS + LangChain 

Module 3: Generative AI on Cloud 

• Choose from leading Foundation models 

• AI21: Jamba, Jurassic 
• Amazon: Titan Models 
• Anthropic: Claude Models 
• Cohere: Command Models 
• Meta: Llama 
• Mistral AI 
• Stability AI Models 
• Open AI Models 
• DeepSeek 
• Google Gemini models 
• Hugging Face models 

• Model Hyper parameter Configurations 
• AWS for Generative AI 

• Getting started with Amazon Bedrock 

• Experiment with Foundation models for different tasks 
• Chat/text playground 
• Image playground 

• Privately customize FMs with your data 
• Amazon Bedrock Converse API 
• Amazon Q – Generative AI Assistant

• Amazon Q Business 
• Amazon Q Developer 

• Amazon Sage Maker for Generative AI 

• Sage Maker JumpStart pre-trained models
• Google AI Studio and Gemini API 

• Google AI Studio Introduction 
• Gemini API Overview 
• Google Ai Studio playground

• AI Playground Chat Audio Docs & Images 
• Real-Time Streaming Audio & Video 

• Vertex AI Studio 

• Vertex AI Studio Getting Started 
• Real-time Media Studio & Streaming 
• Prompt Management Gallery & Optimization 
• Model Tuning & Customization 

• Agent Builder 

• Agent Garden 
• Agent Engine 
• RAG Engine 
• Vertex AI Search 
• Vector Search 

• Vertex AI Model Garden Foundation Models 

• Azure for Generative AI: Azure AI Foundry  Getting started with Azure AI Foundry
• Understanding RBAC Roles in Azure AI Foundry Understanding Azure AI Foundry resources
• AI project 
• AI hub 
• AI Services
• Azure OpenAI Service 

• Azure AI Model Catalog Discover and Deploy AI Models 
• AI Playground Experiment, Customize & Build 
• Azure AI Agent Secure & Scalable Enterprise Automation 
• Fine-Tune AI Models with Your Data 
• Prompt Flow Build & Refine AI Workflows 
• Tracing & Evaluation Debug and Optimize AI Performance 
• AI Safety & Security Build with Confidence 

Module 3.1: Text generation on AWS and GCP 

• Amazon Bedrock Text Generation 

• Hands-on:  Text Generation 

• Leverage Amazon Bedrock to generate high-quality and contextually relevant text. 

• Hands-on: Bedrock model for code generation 

• Using Claude models for code generation 

• Hands-on:  Text Summarization 

• Utilize Titan and Claude models to distill complex information into concise summaries. 

• Hands-on: Question Answering (QnA) 

• Build intelligent QnA systems with the capabilities of the Titan model. 

• Hands-on: Entity Extraction 

• Master advanced techniques for extracting critical entities from text. 
• Azure AI Foundry text generation 

• • Azure Authentication & Environment Setup 
  • Understanding AI Project Client 
  • Azure AI Foundry Quick Start Guide 

• Project: Chat Completions with AI Project Client 

• Project: Getting started with Text Embeddings models 

• AI Foundry Prompt Template 
• Phi-4 Model with AI Project Client 
• Hands-on: Building Advanced Chat Systems with Phi-4
• Azure OpenAI Studio 

• Azure OpenAI Chat on Private Data with LangChain 
• Azure OpenAI Q&A with Semantic Search Using LlamaIndex
• Google AI Studio text generation 

• Text Generation from Text-Only Inputs 
• Generate Content from Combined Text & Images 
• Real-Time Text Streaming 
• Hands-on: Build Interactive Chat Experiences 
• Handling Long Contexts with Gemini 
• Executing Code with Gemini Basics 
• Producing Structured Responses with the Gemini API 
• Gemini 2.0 Rapid Reasoning & Multi-Turn Dialogues 
• Live Multimodal API Implementation 
• Introduction to Function Execution with the Gemini API 

Module 3.2: RAG with AWS, Azure and GCP 

• Hands-on: Amazon Bedrock Knowledge Bases and RAG ○ Managed RAG 

• Retrieve and generate using managed RAG services.
• LangChain RAG 
• Implement RAG workflows using LangChain for retrieval and generation. 

• Hands-on: Azure AI Foundry RAG 

• Azure AI Search for RAG-Powered Applications 
• Embeddings Model for RAG-Based Architecture 
• Embedding, Storing & Chatting with Docs in Azure AI Search 
• Bing Grounding: Enhance AI with Web Search Context

• Hands-on: Vertex AI & Google AI Studio for RAG Solutions 

• RAG architecture using Vertex AI 
• Google Search Grounding: Enrich AI with Real-Time Context 
• Enhance AI with Google Search Suggestions 
• Vertex AI RAG Engine 

Module 4: Image generation and multimodal models 

• Bedrock Titan Image Generator 

• Generate high-quality images using Bedrock Titan.

• Project: Bedrock Amazon Nova 

• Create detailed videos with the power of Amazon Nova Foundation Models 

• Project: Bedrock Titan Multimodal Embeddings 

• Leverage Titan Multimodal embeddings for advanced multimodal AI tasks. 

• Google AI Studio 

• • Imagen 3 in Gemini API 
  • Imagen Model Parameters Overview 
  • Handling Image & Base64 Inputs with Gemini 
  • Video & Text Prompting: Transcription & Visual Descriptions 
  • Google’s Veo 2 and Veo 3 

• Project: Getting started with Google’s Veo 2 

• Azure AI Foundry 

• • Azure AI Foundry Image Generation Capabilities 
  • Embed Images with Azure AI Foundry 

• Project: Getting Started with OpenAI DALL·E 3 for Image Generation 

Module 5: Customizing models via fine tuning 

• Model Customization Techniques in Amazon Bedrock 

• Fine-tuning options in Amazon Bedrock. 
• Data preparation 
• Customizing hyper parameters 
• Fine-Tuning & Retrieve Custom Model 
• Invoke Custom Model 

• Project: Fine-Tuning with the Gemini API 

• Fine-Tuning Process in Google AI Studio 
• Advanced Tuning Settings with the Gemini API 

• Fine-tune models with Azure AI Foundry 
• LoRA (Low-Rank Adaptation) and QLoRA for parameter-efficient tuning 
• PEFT (Parameter-Efficient Fine-Tuning) techniques 

Module 6: Ai Agents 

• • Introduction AI Agents 
  • The Importance of AI Agents 
  • Applications and Use Cases of AI Agents
  • Understanding the workflow of AI agents
  • What are AI agents made of? 

• Project: Agents for Amazon Bedrock (AWS)

• Components of Bedrock Agents 
• Foundation model 
• Instructions 
• Action groups 
• Knowledge bases for AI Agents 
• Guardrails for Amazon Bedrock 
• Getting started with Amazon Bedrock Agents 

• Building AI Agents using Gemini API (GCP) 

• Project: Building AI Workflows with LangChain 

• Creating an automated essay-writing pipeline 
• Integrating LLMs, prompts, and web search 
• Using LLM Chain, Chat Prompt Template, and StrOutputParser 
• Implementing LangChain Expression Language (LCEL) for task automation 
• Implementing structured data validation with Pydantic 
• Using LLM Chain, Chat Prompt Template, and StrOutput Parser 
• Handling real-time information retrieval using Tavily API 
• Iterative AI Agent with LangGraph
• Deploying and Scaling AI Agents 
• Debugging AI agent issues in LangGraph and LangChain 
• Deploying AI Agents on Vertex AI 

• Project: Azure AI Agents 

• Building AI Agents with Azure 
• Understanding AIProjectClient for managing AI workflows 
• Managing AI Conversations 
• Enhancing AI Agents with Tools 
• Code Interpreter Tool: Performing calculations and analyzing datasets 
• File Search Tool: Searching documents and extracting insights 
• Bing Grounding Tool: Fetching real-time search results 
• Azure AI Search Tool: Connecting AI Agents to a structured search index 
• Using AI for Data Search and Retrieval 
• Setting up Azure AI Search for indexing documents 
• Creating and managing Vector Stores for AI-driven search 
• AI Search Integration for Real-time Information 
• Deploying AI Agents in Production 
• Advanced AI Agent Customization

• Project: Google Agent Development Kit

• Core Agent Categories 

• LLM Agents 
• Workflow Agents 
• Custom Agents 
• Multi-Agent Systems in ADK ○ ADK Tool 
• Function Tools 
• Built-in Tools 
• Third-Party Tools 

• Integrating Model Context Protocol (MCP) with ADK 
• Deploying Your Agent 

• Project: Agent2Agent (A2A) 

• Benefits of Using A2A 

• Key Design Principles of A2A 
• The A2A Solution 
• A2A and MCP 
• How A2A and MCP Work Together 
• Agent Discovery in A2A 

• Project: Model Context Protocol (MCP) 

• The Architecture of MCP 

• MCP Servers 
• MCP Clients 
• MCP Hosts 
• Local Data Sources 
• Remote Services 

• MCP Ecosystem and Adoption 

• Project: Introduction to Multi-Agent Systems 

• What are Multi-AI Agents? 
• Real-world Applications of Multi-Agent Systems 
• Overview of LangGraph for Multi-Agent Orchestration 
• Benefits of Multi-Agent Architecture 

• Other Agent framework Auto Gen, CrewAI and Semantic kernel 

Module 7: Monitoring and performance Evaluation 

• Project: Observability in Azure AI foundry 

• Project: Cloud-Based Model Evaluation using AIProjectClient 

• • Observability & Tracing in Azure AI 
  • Azure Monitor & Application Insights 
  • Best Practices for AI Model Monitoring 

• Project: LLM Evaluation & Testing with Evidently AI 

• Evaluate and test your LLM use case 
• Create and evaluate an LLM judge
• Run regression testing for LLM outputs.

• Project: MLFlow for LLM Evaluation 

• Model Evaluation in MLFlow 
• Heuristic-Based Evaluation Metrics 
• LLM-as-a-Judge Evaluation Metrics 
• Custom LLM Evaluation Metrics 

• AWS Bedrock for LLM and RAG Evaluation 
• BLEU, ROUGE, METEOR, and BERT Score for assessing text quality
• Evaluate AI Agents 

• Preparing for Agent Evaluations 
• How Evaluation works with the ADK 

• Retrieval-augmented Generation (RAG) evaluation 

• Evaluating Retrieval Systems on cloud 
• RAG Evaluation with MLflow 

Module 8: Ethics and Deployment in Generative AI 

• Responsible AI 

• What is responsible AI 
• Challenges of responsible AI 
• Amazon services and tools for responsible AI 
• Building AI responsibly at AWS 
• Core dimensions of responsible AI 
• Project: Implementing safeguards in generative AI 

• Amazon Bedrock Guardrails 

• Azure Responsible AI capabilities 
• Google Cloud’s approach to responsible AI 

• MLOps for LLM’s (LLMOps) 

• What is LLM? 
• MLOps for LLM’s 
• FMOps/LLMOps: Operationalize generative AI 
• LLM System Design 
• High-level view LLM-driven application 
• LLMOps Pipeline 

• Enterprise-Ready Features for A2A Agents 

• Transport Level Security (TLS) 
• Authentication 
• Authorization 
• Data Privacy and Confidentiality 
• Tracing, Observability, and Monitoring 
• API Management and Governance 

Module 9: Developer focussed AI tools 

• Project: Amazon Q for Developer 

• Setting Up Amazon Q Developer 
• Conversing with Amazon Q for Code Assistance 
• Inline Code Suggestions with Amazon Q 
• Code Transformation in the IDE Using Amazon Q 
• Feature Development with Amazon Q 
• Automated Unit Test Generation with Amazon Q 
• Code Review with Amazon Q Developer 
• Auto-Generated Documentation with Amazon Q 
• Supported Languages for Amazon Q in IDE 

• Project: GitHub Copilot 

• Best Practices for GitHub Copilot 
• Automating Tests & Repetitive Code with Copilot 
• Debugging & Fixing Syntax Errors Using Copilot 
• Generating Code Explanations & Comments with Copilot 
• Creating Regular Expressions with Copilot 
• AI-Powered Code Completions with GitHub Copilot
• Enhancing Code Reviews with GitHub Copilot 
• Streamlining Pull Requests with Copilot Assistance 

• Other developer focused AI tools 

• Project: Cursor 

• Windsurf 
• Replit