ABOUT THE AUTHORsShreyas Subramanian, PhD, is Principal Machine Learning specialist at Amazon Web Services. He has worked with several enterprise companies on business-critical machine learning and optimization problems. Stefan Natu is Principal Machine Learning Specialist at Alexa AI, prior to which he was a Principal Architect at Amazon Web Services. His professional focus is on financial services, and he helps customers architect ML use cases on AWS with an emphasis on security, enterprise model governance, and operationalizing machine learning models.

• Introduction xviiAssessment Test xxixAnswers to Assessment Test xxxvPart I Introduction 1Chapter 1 AWS AI ML Stack 3Amazon Rekognition 4Image and Video Operations 6Amazon Textract 10Sync and Async APIs 11Amazon Transcribe 13Transcribe Features 13Transcribe Medical 14Amazon Translate 15Amazon Translate Features 16Amazon Polly 17Amazon Lex 19Lex Concepts 19Amazon Kendra 21How Kendra Works 22Amazon Personalize 23Amazon Forecast 27Forecasting Metrics 30Amazon Comprehend 32Amazon CodeGuru 33Amazon Augmented AI 34Amazon SageMaker 35Analyzing and Preprocessing Data 36Training 39Model Inference 40AWS Machine Learning Devices 42Summary 43Exam Essentials 43Review Questions 44Chapter 2 Supporting Services from the AWS Stack 49Storage 50Amazon S3 50Amazon EFS 52Amazon FSx for Lustre 52Data Versioning 53Amazon VPC 54AWS Lambda 56AWS Step Functions 59AWS RoboMaker 60Summary 62Exam Essentials 62Review Questions 63Part II Phases of Machine Learning Workloads 67Chapter 3 Business Understanding 69Phases of ML Workloads 70Business Problem Identification 71Summary 72Exam Essentials 73Review Questions 74Chapter 4 Framing a Machine Learning Problem 77ML Problem Framing 78Recommended Practices 80Summary 81Exam Essentials 81Review Questions 82Chapter 5 Data Collection 85Basic Data Concepts 86Data Repositories 88Data Migration to AWS 89Batch Data Collection 89Streaming Data Collection 92Summary 96Exam Essentials 96Review Questions 98Chapter 6 Data Preparation 101Data Preparation Tools 102SageMaker Ground Truth 102Amazon EMR 104Amazon SageMaker Processing 105AWS Glue 105Amazon Athena 107Redshift Spectrum 107Summary 107Exam Essentials 107Review Questions 109Chapter 7 Feature Engineering 113Feature Engineering Concepts 114Feature Engineering for Tabular Data 114Feature Engineering for Unstructured and Time Series Data 119Feature Engineering Tools on AWS 120Summary 121Exam Essentials 121Review Questions 123Chapter 8 Model Training 127Common ML Algorithms 128Supervised Machine Learning 129Textual Data 138Image Analysis 141Unsupervised Machine Learning 142Reinforcement Learning 146Local Training and Testing 147Remote Training 149Distributed Training 150Monitoring Training Jobs 154Amazon CloudWatch 155AWS CloudTrail 155Amazon Event Bridge 158Debugging Training Jobs 158Hyperparameter Optimization 159Summary 162Exam Essentials 162Review Questions 164Chapter 9 Model Evaluation 167Experiment Management 168Metrics and Visualization 169Metrics in AWS AI/ML Services 173Summary 174Exam Essentials 175Review Questions 176Chapter 10 Model Deployment and Inference 181Deployment for AI Services 182Deployment for Amazon SageMaker 184SageMaker Hosting: Under the Hood 184Advanced Deployment Topics 187Autoscaling Endpoints 187Deployment Strategies 188Testing Strategies 190Summary 191Exam Essentials 191Review Questions 192Chapter 11 Application Integration 195Integration with On-PremisesSystems 196Integration with Cloud Systems 198Integration with Front-EndSystems 200Summary 200Exam Essentials 201Review Questions 202Part III Machine Learning Well-Architected Lens 205Chapter 12 Operational Excellence Pillar for ML 207Operational Excellence on AWS 208Everything as Code 209Continuous Integration and Continuous Delivery 210Continuous Monitoring 213Continuous Improvement 214Summary 215Exam Essentials 215Review Questions 217Chapter 13 Security Pillar 221Security and AWS 222Data Protection 223Isolation of Compute 224Fine-GrainedAccess Controls 225Audit and Logging 226Compliance Scope 227Secure SageMaker Environments 228Authentication and Authorization 228Data Protection 231Network Isolation 232Logging and Monitoring 233Compliance Scope 235AI Services Security 235Summary 236Exam Essentials 236Review Questions 238Chapter 14 Reliability Pillar 241Reliability on AWS 242Change Management for ML 242Failure Management for ML 245Summary 246Exam Essentials 246Review Questions 247Chapter 15 Performance Efficiency Pillar for ML 251Performance Efficiency for ML on AWS 252Selection 253Review 254Monitoring 255Trade-offs256Summary 257Exam Essentials 257Review Questions 258Chapter 16 Cost Optimization Pillar for ML 261Common Design Principles 262Cost Optimization for ML Workloads 263Design Principles 263Common Cost Optimization Strategies 264Summary 266Exam Essentials 266Review Questions 267Chapter 17 Recent Updates in the AWS AI/ML Stack 271New Services and Features Related to AI Services 272New Services 272New Features of Existing Services 275New Features Related to Amazon SageMaker 279Amazon SageMaker Studio 279Amazon SageMaker Data Wrangler 279Amazon SageMaker Feature Store 280Amazon SageMaker Clarify 281Amazon SageMaker Autopilot 282Amazon SageMaker JumpStart 283Amazon SageMaker Debugger 283Amazon SageMaker Distributed Training Libraries 284Amazon SageMaker Pipelines and Projects 284Amazon SageMaker Model Monitor 284Amazon SageMaker Edge Manager 285Amazon SageMaker Asynchronous Inference 285Summary 285Exam Essentials 285Appendix Answers to the Review Questions 287Chapter 1: AWS AI ML Stack 288Chapter 2: Supporting Services from the AWS Stack 289Chapter 3: Business Understanding 290Chapter 4: Framing a Machine Learning Problem 291Chapter 5: Data Collection 291Chapter 6: Data Preparation 292Chapter 7: Feature Engineering 293Chapter 8: Model Training 294Chapter 9: Model Evaluation 295Chapter 10: Model Deployment and Inference 295Chapter 11: Application Integration 296Chapter 12: Operational Excellence Pillar for ML 297Chapter 13: Security Pillar 298Chapter 14: Reliability Pillar 298Chapter 15: Performance Efficiency Pillar for ML 299Chapter 16: Cost Optimization Pillar for ML 300Index 303