book

A Practical Guide to Generative AI Using Amazon Bedrock: Building, Deploying, and Securing Generative AI Applications

Name: A Practical Guide to Generative AI Using Amazon Bedrock: Building, Deploying, and Securing Generative AI Applications
Author: Avik Bhattacharjee
ISBN: 9798868814143

by Avik Bhattacharjee

July 2025

Intermediate to advanced

489 pages

9h 23m

English

Apress

Read now

Unlock full access

A Practical Guide to Generative AI Using Amazon Bedrock
Acknowledgments
Table of Contents
About the Author
About the Technical Reviewer
1. Introduction to Generative AI
1.1 Understanding Generative AIExploring the Bright Future of Cutting-Edge TechnologyIntroduction to Artificial Intelligence (AI)Human Vision Schematic FlowArtificial Intelligence Computer Vision Schematic FlowIntroduction to Generative AIA Comprehensive Taxonomy and ClassificationUnderlying Principles and ArchitecturesTransformer-Based ModelsEncoder TransformerPositional EncodingSelf-Attention MechanismA. Multi-head Attention MechanismB. Feed-Forward Neural NetworkDecoder TransformerA. The Masked Multi-head AttentionB. The Multi-head AttentionC. The Feed-Forward Neural NetworkVariation of the Transformer ArchitectureEncoder OnlyDecoder OnlyEncoder-Decoder OnlyThe Transformer Encoder’s DrawbacksThe Transformer Decoder’s DrawbacksImportance and Applications1.2 The Infinite Potential of Generative AI for the Future1.3 Summary
2. Generative AI with AWS
2.1 AWS Generative AI StackBottom Layer of the Stack: Infrastructure for FM Training and InferencesAmazon SageMaker AIMiddle Layer of the Stack: Tools to Build with LLMs and Other FMsTop Layer of the Stack: Applications That Leverage LLMs and Other FMsAmazon Q2.2 Potential Industry Use Cases for Generative AI2.3 Why Generative AI on AWS2.4 Accelerating Generative AI Application Development on AWS2.5 Generative AI Project Lifecycle2.6 Summary
3. Introduction to Amazon Bedrock
3.1 What Is Amazon Bedrock3.2 Features of Amazon Bedrock3.3 Set Up Amazon BedrockConsole Access3.4 Foundation Models on Amazon BedrockModel ProviderModel IDModel FamilyModel VersionModel Input ModalitiesModel Output ModalitiesModel Size3.5 Model Lifecycle3.6 Amazon Bedrock Console Walk-ThroughGetting StartedFoundation ModelsPlaygroundsBuilder ToolsSafeguardsInference and AssessmentModel AccessModel Invocation Logging3.7 Running Model InferencesInference ParametersRandomness and DiversityTemperatureTop KTop PCombined Impact of Inference ParametersLengthImage Playground3.8 Amazon SageMaker and Amazon Bedrock Interaction3.9 Summary
4. Overview of Prompt Engineering and In-Context Learning
4.1 What Are Prompts4.2 What Is Prompt Engineering4.3 Components of Prompts4.4 Introduction to Text, Token, and Embedding4.5 Prompt Templates4.6 Overview of In-Context Learning4.7 Various In-Context Learning MethodsZero-Shot Learning (ZSL)One-Shot Learning (OSL)Few-Shot Learning (FSL)Chain of Thought (CoT)4.8 Guidelines of Prompt Engineering and In-Context Learning on Amazon Bedrock4.9 Converse API4.10 Sample Application: Using Converse API4.11 Summary
5. Overview of Use Cases in This Book
5.1 Q&A: Enhanced Prompt for Knowledge Search5.2 Virtual Assistance: Safeguard Generated Response5.3 Application: Orchestrate Series of Actions or Steps5.4 Financial Virtual Assistance: Relevant Information5.5 Summary

6. Overview of Retrieval-Augmented Generation (RAG)
6.1 Introduction to RAG6.2 Understanding Retrieval Models6.3 Why Retrieval-Augmented Generation6.4 RAG ArchitectureData Preparation and OrganizationExperiences, Information Retrieval, and Generation6.5 Overview of Embeddings6.6 Overview of LangChain6.7 Overview of LlamaIndex6.8 Overview of a Simple Streamlit Application6.9 A Sample Application Building with RAG6.10 Challenges and Considerations6.11 Advanced RAG Design PatternsFundamental RAGHyDE RAGMulti-query RAGSentence Window Retrieval RAGDocument Summary Index RAGReranker RAGT-RAGAgentic RAG6.12 Summary
7. Overview of Amazon Bedrock Knowledge Bases
7.1 Introduction to Amazon Bedrock Knowledge BasesData Preparation and OrganizationExperiences and Information Retrieval and Generation7.2 Why Amazon Bedrock Knowledge Bases7.3 Sample Applications of Amazon Bedrock Knowledge Bases7.4 Overview of Chunking StrategyFixed-Size ChunkingNo ChunkingHierarchical ChunkingSemantic ChunkingCustom ChunkingChoosing the Right Strategy7.5 Governance and Monitoring7.6 Design Principles of Right Vector DB7.7 Summary
8. Overview of Safeguard’s Practice
8.1 Introduction to Responsible AI8.2 Why Responsible AI Is Important8.3 Introduction to Amazon Bedrock GuardrailsProperties of Amazon Bedrock GuardrailsUnderstanding Amazon Guardrails: A Functional Block DiagramComponents of a GuardrailContent FiltersDenied TopicsSensitive Information FiltersWord FiltersContextual Grounding Check8.4 Sample Application: Building Amazon Bedrock Guardrails8.5 Introduction to Watermark Detection8.6 Understanding the Importance of Watermark Detection8.7 Governance and Monitoring8.8 Summary
9. Overview of Amazon Bedrock Agents
9.1 Introduction to Amazon Bedrock Agents9.2 Why Amazon Bedrock AgentsDynamic Tool OrchestrationContext-Aware ResponsesTask Automation for EfficiencyScalability and Modular Design9.3 Simple Applications of Amazon Bedrock Agents9.4 Challenges and Considerations9.5 Governance and MonitoringMonitoring ComponentsGovernance Framework9.6 Summary
10. Overview of Model Customization
10.1 Introduction to Fine-TuningSteps in Fine-Tuning10.2 Overview of Important Terms for Fine-TuningParameter-Efficient Fine-Tuning (PEFT)Low-Rank Adaptation (LoRA)Hyperparameters10.3 Why Fine-Tuning10.4 Comparison: In-Context Learning, Full Training, and Fine-Tuning10.5 Introduction to Continuous Pre-training10.6 Why Continuous Pre-training10.7 Comparison: In-Context Learning, Full Training, and Continuous Pre-training10.8 Simple Applications of Amazon Bedrock Model Customization10.9 Comparison: Fine-Tuning and Continuous Pre-training10.10 Strategy to Choose Between Fine-Tuning and Continuous Pre-training10.11 Comparison Between RAG and Model Customization10.12 Strategy to Choose Between RAG and Model CustomizationFinal Decision Framework10.13 Monitoring and Governance10.14 Summary
11. Overview of Model Evaluation
11.1 Introduction to Model Evaluation11.2 Model Evaluation on Amazon Bedrock11.3 Automatic Model Evaluation JobAccuracyBERTScoreRobustnessClassification Accuracy ScoreDelta Classification Accuracy ScoreToxicity11.4 Human Model Evaluation Job11.5 Overview of LLM As a Judge11.6 Overview of Knowledge Base Evaluation11.7 Sample Application of Model Evaluation on Amazon Bedrock11.8 Summary
12. Overview of Best Model Selection and Best Practices
12.1 Introduction to the Cost-Quality-Latency Triangle12.2 Understanding the PricingUse Case: Virtual Assistance with In-Context LearningUse Case: Virtual Assistance with RAG Design PatternsUse Case: Virtual Assistance with Fine-TuningUse Case: Virtual Image Gallery with In-Context LearningUse Case: Virtual Image Gallery with Fine-Tuning12.3 Strategy and Best Practices to Build Complete SolutionUnderstanding Business Problems Before Jumping to Generative AIChoosing the Right Foundation Model for Business SolutionsLarge Foundation Models Are Not Always the Right SolutionMental Model for Different Design PatternsBuilding Unique Knowledge Management Framework to Boost RAG Design PatternsAgentic Solutions vs. Simplicity: When Straightforward Workflows Outperform ComplexityTypes of Workflows and Their Use CasesWorkflow with ChainingWorkflow with RoutingWorkflow with Parallelization12.4 Summary
13. Overview of Security and Privacy of Amazon Bedrock
13.1 Data Protection13.2 Identity and Access Management for Amazon BedrockIdentity-Based PoliciesAWS Managed Policies13.3 Compliance Validation for Amazon Bedrock13.4 Incident Response in Amazon Bedrock13.5 Resilience in Amazon Bedrock13.6 Infrastructure Security in Amazon Bedrock13.7 Using Interface VPC Endpoints (AWS PrivateLink)13.8 Prompt Injection Security13.9 Generative AI Security Scoping Matrix13.10 Summary
14. Overview of GenAIOPS
14.1 Introduction to MLOps14.2 Introduction to MLOps Evolution Framework14.3 Introduction to GenAIOps14.4 Challenges and Considerations in GenAIOps14.5 Enterprise-Scale GenAIOps Application: A Step-by-Step Layer14.6 Impact of People, Process, and Technology14.7 Summary
15. Overview of Prompt Management
15.1 Overview of Prompt Management15.2 Lifecycle of Prompt ManagementPrompt DesignTesting and Baseline EvaluationIterative RefinementVersion ControlMonitoring and Continuous FeedbackDynamic AdaptationScaling and ReusabilityRetirement or Deprecation15.3 Sample Application: Building Amazon Bedrock Prompt Management15.4 Creating a Prompt15.5 Testing a Prompt15.6 Managing a Prompt15.7 Deploying a Prompt15.8 Summary
16. Overview of Prompt Flow
16.1 Introduction to Prompt Flow or Amazon Bedrock FlowKey ComponentsStrategyImportant Terms16.2 Design Patterns of Prompt Flow or Amazon Bedrock FlowMulti-LLM Routing PatternChain-of-Thought (CoT) Reasoning PatternPrompt Chaining PatternFew-Shot Learning Pattern16.3 Sample Application: Building Prompt Flow or Amazon Bedrock Flow16.4 Summary
17. Overview of Provisioned Throughput
17.1 Introduction to Provisioned Throughput17.2 Why Provisioned Throughput Is Important17.3 Overview of Model Units (MUs)17.4 Challenges and Considerations of Provisioned Throughput17.5 Simple Applications of Amazon Bedrock Provisioned Throughput17.6 Summary
18. Overview of Image Capabilities
18.1 Introduction to Generative AI for Image Creation18.2 Understanding the Functioning of Image on Generative AI18.3 Progress and Innovations in Image on Generative AI18.4 Addressing Challenges in Image on Generative AIGuaranteeing Image Quality and RealismMitigating Bias in Generated ContentScaling Image Generation EfficientlySafeguarding Intellectual Property and Ethical Concerns18.5 Exploring the Ethical Dimensions of Image on Generative AIContent Authenticity and MisinformationRepresentation BiasIntellectual Property and OwnershipEnvironmental Impact of AI Image Generation18.6 Advanced Patterns of Image on Generative AIPerfecting Prompt for ImageImage EmbeddingImage to ImageImage InpaintingImage ConditioningColor ConditioningImage OutpaintingBackground RemovalCombination of Text and ImageImage Model Customization18.7 Sample Application with Image on Generative AI18.8 Summary
19. Overview of Multimodal Capabilities
19.1 Introduction to Multimodal Generative AI19.2 Understanding the Functioning of Multimodal Generative AI19.3 Progress and Innovations in Multimodal Generative Models19.4 Addressing Challenges in Multimodal Generative AIChallenges in Modality Fusion AlignmentHandling Noisy and Incomplete DatasetsScaling Multimodal Models EfficientlyEthical Challenges in Multimodal Generative AI19.5 Exploring the Ethical Dimensions of Multimodal CapabilitiesEnsuring Fairness in Multimodal CapabilitiesRisks of Multimodal Deepfakes and MisuseBalancing Innovation and Accountability in AI19.6 Sample Application with Multimodal Capability ArchitectureVector DB CreationSearch Capability with a Simple Text PromptSearch Capability Featuring a Combination of Text and Image Prompts19.7 Sample Application with Multimodal Capability19.8 Summary
20. Conclusion
20.1 Recap of Key Concepts20.2 Looking Ahead: Future Developments in Generative AI20.3 Summary
Index

Content preview from A Practical Guide to Generative AI Using Amazon Bedrock: Building, Deploying, and Securing Generative AI Applications

A. BhattacharjeeA Practical Guide to Generative AI Using Amazon Bedrockhttps://doi.org/10.1007/979-8-8688-1414-3_11

11. Overview of Model Evaluation

Avik Bhattacharjee¹

In this chapter, you will learn about ...

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Modern Automated AI Agents: Building Agentic AI to Perform Complex Tasks

Publisher Resources

ISBN: 9798868814143Purchase Link Publisher Website

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

A Practical Guide to Generative AI Using Amazon Bedrock: Building, Deploying, and Securing Generative AI Applications

by Avik Bhattacharjee

11. Overview of Model Evaluation

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.