{"product_id":"machine-learning-in-production-isbn-9780262049726","title":"Machine Learning in Production","description":"\u003cb\u003eA practical and innovative textbook detailing how to build real-world software products with machine learning components, not just models.\u003cbr\u003e\u003c\/b\u003e\u003cbr\u003e\u003cbr\u003eTraditional machine learning texts focus on how to train and evaluate the machine learning model, while MLOps books focus on how to streamline model development and deployment. But neither focus on how to build actual products that deliver value to users. This practical textbook, by contrast, details how to responsibly build products with machine learning components, covering the entire development lifecycle from requirements and design to quality assurance and operations. Machine Learning in Production brings an engineering mindset to the challenge of building systems that are usable, reliable, scalable, and safe within the context of real-world conditions of uncertainty, incomplete information, and resource constraints. Based on the author’s popular class at Carnegie Mellon, this pioneering book integrates foundational knowledge in software engineering and machine learning to provide the holistic view needed to create not only prototype models but production-ready systems. \u003cbr\u003e\u003cbr\u003e•   Integrates coverage of cutting-edge research, existing tools, and real-world applications\u003cbr\u003e•   Provides students and professionals with an engineering view for production-ready machine learning systems\u003cbr\u003e•   Proven in the classroom\u003cbr\u003e•   Offers supplemental resources including slides, videos, exams, and further readingsI SETTING THE STAGE 2\u003cbr\u003e1 Introduction 3\u003cbr\u003e1.1 Motivating Example: An Automated Transcription Startup 4\u003cbr\u003e1.2 Data Scientists and Software Engineers 6\u003cbr\u003e1.3 Machine-Learning Challenges in Software Projects 8\u003cbr\u003e1.4 A Foundation for MLOps and Responsible Engineering 13\u003cbr\u003e1.5 Summary 15\u003cbr\u003e1.6 Further Readings 16\u003cbr\u003e2 From Models to Systems 19\u003cbr\u003e2.1 ML and Non-ML Components in a System 19\u003cbr\u003e2.2 Beyond the Model 24\u003cbr\u003e2.3 On Terminology 29\u003cbr\u003e2.4 Summary 30\u003cbr\u003e2.5 Further Readings 30\u003cbr\u003e3 Machine Learning for Software Engineers, in a Nutshell 33\u003cbr\u003e3.1 Basic Terms: Machine Learning, Models, Predictions 33\u003cbr\u003e3.2 Technical Concepts: Model Parameters, Hyperparameters, Model Storage 34\u003cbr\u003e3.3 Machine Learning Pipelines 35\u003cbr\u003e3.4 Foundation Models and Prompting 36\u003cbr\u003e3.5 On Terminology 37\u003cbr\u003e3.6 Summary 38\u003cbr\u003e3.7 Further Readings 38\u003cbr\u003eII REQUIREMENTS ENGINEERING 39\u003cbr\u003e4 When to use Machine Learning 41\u003cbr\u003e4.1 Problems that Benefit from Machine Learning 41\u003cbr\u003e4.2 Tolerating Mistakes and ML Risk 42\u003cbr\u003e4.3 Continuous Learning 43\u003cbr\u003e4.4 Costs and Benefits 43\u003cbr\u003e4.5 The Business Case: Machine Learning as Predictions 44\u003cbr\u003e4.6 Summary 45\u003cbr\u003e4.7 Further Readings 45\u003cbr\u003e5 Setting and Measuring Goals 47\u003cbr\u003e5.1 Scenario: Self-help legal chatbot 47\u003cbr\u003e5.2 Setting Goals 48\u003cbr\u003e5.3 Measurement in a Nutshell 51\u003cbr\u003e5.4 Summary 57\u003cbr\u003e5.5 Further Readings 57\u003cbr\u003e6 Gathering Requirements 59\u003cbr\u003e6.1 Scenario: Fall Detection with a Smart Watch 60\u003cbr\u003e6.2 Untangling Requirements 60\u003cbr\u003e6.3 Eliciting Requirements 66\u003cbr\u003e6.4 How Much Requirements Engineering and When? 71\u003cbr\u003e6.5 Summary 72\u003cbr\u003e6.6 Further Readings 73\u003cbr\u003e7 Planning for Mistakes 75\u003cbr\u003e7.1 Mistakes Will Happen 76\u003cbr\u003e7.2 Designing for Failures 78\u003cbr\u003e7.3 Hazard Analysis and Risk Analysis 84\u003cbr\u003e7.4 Summary 89\u003cbr\u003e7.5 Further Readings 90\u003cbr\u003eIII ARCHITECTURE AND DESIGN 92\u003cbr\u003e8 Thinking like a Software Architect 93\u003cbr\u003e8.1 Quality Requirements Drive Architecture Design 94\u003cbr\u003e8.2 The Role of Abstraction 97\u003cbr\u003e8.3 Common Architectural Design Challenges for ML-Enabled Systems 97\u003cbr\u003e8.4 Codifying Design Knowledge 100\u003cbr\u003e8.5 Summary 105\u003cbr\u003e8.6 Further Readings 105\u003cbr\u003e9 Quality Attributes of ML Components 109\u003cbr\u003e9.1 Scenario: Detecting Credit Card Fraud 109\u003cbr\u003e9.2 From System Quality to Model and Pipeline Quality 109\u003cbr\u003e9.3 Common Quality Attributes 111\u003cbr\u003e9.4 Constraints and Tradeoffs 115\u003cbr\u003e9.5 Summary 117\u003cbr\u003e9.6 Further Readings 118\u003cbr\u003e10 Deploying a Model 119\u003cbr\u003e10.1 Scenario: Augmented Reality Translation 119\u003cbr\u003e10.2 Model Inference Function 120\u003cbr\u003e10.3 Feature Encoding 120\u003cbr\u003e10.4 Model Serving Infrastructure 123\u003cbr\u003e10.5 Deployment Architecture Tradeoffs 126\u003cbr\u003e10.6 Model Inference in a System 131\u003cbr\u003e10.7 Documenting Model-Inference Interfaces 135\u003cbr\u003e10.8 Summary 136\u003cbr\u003e10.9 Further Readings 138\u003cbr\u003e11 Automating the Pipeline 141\u003cbr\u003e11.1 Scenario: Home Value Prediction 141\u003cbr\u003e11.2 Supporting Evolution and Experimentation by Designing for Change 142\u003cbr\u003e11.3 Pipeline Thinking 143\u003cbr\u003e11.4 Stages of Machine-Learning Pipelines 144\u003cbr\u003e11.5 Automation and Infrastructure Design 149\u003cbr\u003e11.6 Summary 151\u003cbr\u003e11.7 Further Readings 152\u003cbr\u003e12 Scaling the System 155\u003cbr\u003e12.1 Scenario: Google-Scale Photo Hosting and Search 155\u003cbr\u003e12.2 Scaling by Distributing Work 156\u003cbr\u003e12.3 Data Storage at Scale 157\u003cbr\u003e12.4 Distributed Data Processing 166\u003cbr\u003e12.5 Distributed Machine-Learning Algorithms 176\u003cbr\u003e12.6 Performance Planning and Monitoring 178\u003cbr\u003e12.7 Summary 178\u003cbr\u003e12.8 Further Readings 179\u003cbr\u003e13 Planning for Operations 181\u003cbr\u003e13.1 Scenario: Blogging Platform with Spam Filter 182\u003cbr\u003e13.2 Service Level Objectives 182\u003cbr\u003e13.3 Observability 183\u003cbr\u003e13.4 Automating Deployments 185\u003cbr\u003e13.5 Infrastructure as Code and Virtualization 186\u003cbr\u003e13.6 Orchestrating and Scaling Deployments 188\u003cbr\u003e13.7 Elevating Data Engineering 189\u003cbr\u003e13.8 Incident Response Planning 190\u003cbr\u003e13.9 DevOps and MLOps Principles 191\u003cbr\u003e13.10 DevOps and MLOps Tooling 192\u003cbr\u003e13.11 Summary 195\u003cbr\u003e13.12 Further Readings 195\u003cbr\u003eIV QUALITY ASSURANCE 197\u003cbr\u003e14 Quality Assurance Basics 199\u003cbr\u003e14.1 Testing 200\u003cbr\u003e14.2 Code Review 204\u003cbr\u003e14.3 Static Analysis 205\u003cbr\u003e14.4 Other Quality Assurance Approaches 206\u003cbr\u003e14.5 Planning and Process Integration 207\u003cbr\u003e14.6 Summary 209\u003cbr\u003e14.7 Further Readings 209\u003cbr\u003e15 Model Quality 211\u003cbr\u003e15.1 Scenario: Cancer Prognosis 211\u003cbr\u003e15.2 Defining Correctness and Fit 212\u003cbr\u003e15.3 Measuring Prediction Accuracy 217\u003cbr\u003e15.4 Model Evaluation Beyond Accuracy 231\u003cbr\u003e15.5 Test Data Adequacy 244\u003cbr\u003e15.6 Model Inspection 245\u003cbr\u003e15.7 Summary 245\u003cbr\u003e15.8 Further Readings 246\u003cbr\u003e16 Data Quality 251\u003cbr\u003e16.1 Scenario: Inventory Management 251\u003cbr\u003e16.2 Data Quality Challenges 252\u003cbr\u003e16.3 Data Quality Checks 255\u003cbr\u003e16.4 Drift and Data Quality Monitoring 260\u003cbr\u003e16.5 Data Quality is a System-Wide Concern 264\u003cbr\u003e16.6 Summary 268\u003cbr\u003e16.7 Further Readings 269\u003cbr\u003e17 Pipeline Quality 273\u003cbr\u003e17.1 Silent Mistakes in ML Pipelines 273\u003cbr\u003e17.2 Code Review for ML Pipelines 274\u003cbr\u003e17.3 Testing Pipeline Components 275\u003cbr\u003e17.4 Static Analysis of ML Pipelines 284\u003cbr\u003e17.5 Process Integration and Test Maturity 284\u003cbr\u003e17.6 Summary 285\u003cbr\u003e17.7 Further Readings 286\u003cbr\u003e18 System Quality 287\u003cbr\u003e18.1 Limits of Modular Reasoning 287\u003cbr\u003e18.2 System Testing 289\u003cbr\u003e18.3 Testing Component Interactions and Safeguards 291\u003cbr\u003e18.4 Testing Operations (Deployment, Monitoring) 293\u003cbr\u003e18.5 Summary 293\u003cbr\u003e18.6 Further Readings 294\u003cbr\u003e19 Testing and Experimenting in Production 295\u003cbr\u003e19.1 A Brief History of Testing in Production 295\u003cbr\u003e19.2 Scenario: Meeting Minutes for Video Calls 297\u003cbr\u003e19.3 Measuring System Success in Production 297\u003cbr\u003e19.4 Measuring Model Quality in Production 298\u003cbr\u003e19.5 Designing and Implementing Quality Measures with Telemetry 302\u003cbr\u003e19.6 Experimenting in Production 306\u003cbr\u003e19.7 Summary 311\u003cbr\u003e19.8 Further Readings 312\u003cbr\u003eV PROCESS AND TEAMS 314\u003cbr\u003e20 Data Science and Software Engineering Process Models 315\u003cbr\u003e20.1 Data-Science Process 315\u003cbr\u003e20.2 Software-Engineering Process 318\u003cbr\u003e20.3 Tensions between Data Science and Software Engineering Processes 321\u003cbr\u003e20.4 Integrated Processes for AI-Enabled Systems 323\u003cbr\u003e20.5 Summary 327\u003cbr\u003e20.6 Further Readings 327\u003cbr\u003e21 Interdisciplinary Teams 329\u003cbr\u003e21.1 Scenario: Fighting Depression on Social Media 329\u003cbr\u003e21.2 Unicorns are not Enough 330\u003cbr\u003e21.3 Conflicts Within and Between Teams are Common 331\u003cbr\u003e21.4 Coordination Costs 332\u003cbr\u003e21.5 Conflicting Goals and T-Shaped People 337\u003cbr\u003e21.6 Groupthink 339\u003cbr\u003e21.7 Team Structure and Allocating Experts 340\u003cbr\u003e21.8 Learning from DevOps and MLOps Culture 342\u003cbr\u003e21.9 Summary 345\u003cbr\u003e21.10 Further Readings 346\u003cbr\u003e22 Technical Debt 349\u003cbr\u003e22.1 Scenario: Automated Delivery Robots 349\u003cbr\u003e22.2 Deliberate and Prudent Technical Debt 349\u003cbr\u003e22.3 Technical Debt in Machine Learning Projects 351\u003cbr\u003e22.4 Managing Technical Debt 353\u003cbr\u003e22.5 Summary 354\u003cbr\u003e22.6 Further Readings 355\u003cbr\u003eVI RESPONSIBLE ML ENGINEERING 356\u003cbr\u003e23 Responsible Engineering 357\u003cbr\u003e23.1 Legal and Ethical Responsibilities 357\u003cbr\u003e23.2 Why Responsible Engineering Matters for ML-Enabled Systems 359\u003cbr\u003e23.3 Facets of Responsible ML Engineering 362\u003cbr\u003e23.4 Regulation is Coming 363\u003cbr\u003e23.5 Summary 366\u003cbr\u003e23.6 Further Readings 366\u003cbr\u003e24 Versioning, Provenance, and Reproducibility 369\u003cbr\u003e24.1 Scenario: Debugging a Loan Decision 370\u003cbr\u003e24.2 Versioning 370\u003cbr\u003e24.3 Data Provenance and Lineage 375\u003cbr\u003e24.4 Reproducibility 378\u003cbr\u003e24.5 Putting the Pieces Together 380\u003cbr\u003e24.6 Summary 381\u003cbr\u003e24.7 Further Readings 382\u003cbr\u003e25 Explainability 385\u003cbr\u003e25.1 Scenario: Proprietary Opaque Models for Recidivism Risk Assessment 385\u003cbr\u003e25.2 Defining Explainability 386\u003cbr\u003e25.3 Explaining a Model 389\u003cbr\u003e25.4 Explaining a Prediction 392\u003cbr\u003e25.5 Explaining Data and Training 397\u003cbr\u003e25.6 The Dark Side of Explanations 397\u003cbr\u003e25.7 Summary 398\u003cbr\u003e25.8 Further Readings 398\u003cbr\u003e26 Fairness 401\u003cbr\u003e26.1 Scenario: Mortgage Applications 402\u003cbr\u003e26.2 Fairness Concepts 403\u003cbr\u003e26.3 Measuring and Improving Fairness at the Model Level 410\u003cbr\u003e26.4 Fairness is a System-Wide Concern 416\u003cbr\u003e26.5 Summary 428\u003cbr\u003e26.6 Further Readings 429\u003cbr\u003e27 Safety 433\u003cbr\u003e27.1 Safety and Reliability 433\u003cbr\u003e27.2 Improving Model Reliability 434\u003cbr\u003e27.3 Building Safer Systems 438\u003cbr\u003e27.4 The AI Alignment Problem 442\u003cbr\u003e27.5 Summary 444\u003cbr\u003e27.6 Further Readings 444\u003cbr\u003e28 Security and Privacy 447\u003cbr\u003e28.1 Scenario: Content Moderation 447\u003cbr\u003e28.2 Security Requirements 448\u003cbr\u003e28.3 Attacks and Defenses 449\u003cbr\u003e28.4 ML-Specific Attacks 450\u003cbr\u003e28.5 Threat Modeling 459\u003cbr\u003e28.6 Designing for Security 462\u003cbr\u003e28.7 Data Privacy 466\u003cbr\u003e28.8 Summary 470\u003cbr\u003e28.9 Further Readings 470\u003cbr\u003e29 Transparency and Accountability 473\u003cbr\u003e29.1 Transparency of the Model’s Existence 473\u003cbr\u003e29.2 Transparency of How the Model Works 474\u003cbr\u003e29.3 Human Oversight and Appeals 477\u003cbr\u003e29.4 Accountability and Culpability 478\u003cbr\u003e29.5 Summary 479\u003cbr\u003e29.6 Further Readings 479\u003cb\u003eChristian Kästner \u003c\/b\u003eis associate professor of computer science at Carnegie Mellon University.","brand":"The MIT Press","offers":[{"title":"Default Title","offer_id":46303744295141,"sku":"NP9780262049726","price":85.0,"currency_code":"USD","in_stock":false}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1842\/7735\/files\/9780262049726.jpg?v=1767732013","url":"https:\/\/k12savings.com\/products\/machine-learning-in-production-isbn-9780262049726","provider":"K12savings","version":"1.0","type":"link"}