chiphuyen/dmls-book

Features

• Production Machine Learning Guides - Serves as a comprehensive architecture reference for designing, deploying, and maintaining production-ready ML systems.
• ML Lifecycle Managers - Plans and manages the complete machine learning lifecycle from system design to deployment and maintenance.
• ML Workflow Planners - Maps the key decisions and components needed to develop, deploy, and update machine learning models in production.
• Pipeline Automation Layers - Orchestrates end-to-end ML workflows through configurable DAGs with automated triggers and retry logic.
• Feature Stores - Covers feature store abstraction for standardizing feature computation, storage, and serving across ML teams.
• Machine Learning Systems - Designs reliable, scalable, and maintainable machine learning systems for production deployment and changing business requirements.
• MLOps Best Practices - Documents proven MLOps strategies for automation, monitoring, and shared infrastructure in ML pipelines.
• Model Registries - Describes a centralized version control system for storing, tracking, and managing trained models across environments.
• Model Versioning - Centralizes version control for storing, tracking, and managing trained models across different deployment environments.
• Production Monitoring Frameworks - Provides a structured framework for detecting performance degradation and bias in deployed ML models.
• Production Monitoring Stacks - Provides a continuous observability system that tracks model performance metrics, data drift, and algorithmic bias in real-time.
• ML System Design Patterns - Designs machine learning systems that stay reliable, scalable, and maintainable as business needs change over time.
• ML Lifecycle Pipelines - Organizes the ML process into distinct stages with clear handoff points from data collection to production monitoring.
• DAG-Based Orchestration - Describes DAG-based pipeline orchestration for end-to-end ML workflows with automated triggers and retry logic.
• MLOps Pipeline Automation - Automates model operations and establishes shared infrastructure like model registries and feature stores for reproducible workflows.
• MLOps - Organizes the ML process into distinct stages from data collection to production monitoring with clear handoff points.
• ML Infrastructure Managers - Standardizes shared tools like model stores and feature stores so multiple teams can reuse ML assets across the organization.
• ML System Design References - Provides a reference for system design, workflow automation, and responsible AI implementation across the ML lifecycle.
• ML Production Monitors - Continuously tracks model performance metrics, data drift, and algorithmic bias through an observability system.
• AI Ethics and Fairness - Implements built-in fairness checks and bias detection mechanisms that validate model outputs against ethical guidelines.
• Fairness and Bias Detection Mechanisms - Ships built-in fairness checks and bias detection mechanisms that validate model outputs against ethical guidelines.
• Model Registries - Provides guidance on establishing model registries and feature stores as shared ML infrastructure.
• Bias Detection Guardrails - Validates model outputs against ethical guidelines using built-in fairness checks and bias detection mechanisms.
• Infrastructure as Code - Provides declarative configuration templates for provisioning and managing ML infrastructure components.
• Declarative Infrastructure Provisioning - Uses declarative configuration files to provision and manage ML infrastructure components across cloud environments.

Star history