Harness the power of MLOps for managing real time machine learning project cycle
Key Features
● Comprehensive coverage of MLOps concepts, architecture, tools and techniques.
● Practical focus on building end-to-end ML Systems for Continual Learning with MLOps.
● Actionable insights on CI/CD, monitoring, continual model training and automated retraining.
Description
MLOps, a combination of DevOps, data engineering, and machine learning, is crucial for delivering high-quality machine learning results due to the dynamic nature of machine learning data. This book delves into MLOps, covering its core concepts, components, and architecture, demonstrating how MLOps fosters robust and continuously improving machine learning systems.
By covering the end-to-end machine learning pipeline from data to deployment, the book helps readers implement MLOps workflows. It discusses techniques like feature engineering, model development, A/B testing, and canary deployments. The book equips readers with knowledge of MLOps tools and infrastructure for tasks like model tracking, model governance, metadata management, and pipeline orchestration. Monitoring and maintenance processes to detect model degradation are covered in depth. Readers can gain skills to build efficient CI/CD pipelines, deploy models faster, and make their ML systems more reliable, robust and production-ready.
Overall, the book is an indispensable guide to MLOps and its applications for delivering business value through continuous machine learning and AI.
What you will learn
● Architect robust MLOps infrastructure with components like feature stores.
● Leverage MLOps tools like model registries, metadata stores, pipelines.
● Build CI/CD workflows to deploy models faster and continually.
● Monitor and maintain models in production to detect degradation.
● Create automated workflows for retraining and updating models in production.
Who this book is for
Machine learning specialists, data scientists, DevOps professionals, software development teams, and all those who want to adopt the DevOps approach in their agile machine learning experiments and applications. Prior knowledge of machine learning and Python programming is desired.
Table of Contents
1. Getting Started with MLOps
2. MLOps Architecture and Components
3. MLOps Infrastructure and Tools
4. What are Machine Learning Systems?
5. Data Preparation and Model Development
6. Model Deployment and Serving
7. Continuous Delivery of Machine Learning Models
8. Continual Learning
9. Continuous Monitoring, Logging, and Maintenance
Raman Jhajj is a passionate leader in the data and software engineering space with experience building high-performing teams and leading organizations to become datadriven. He has experience in leading the development of SaaS applications, modern data platforms and MLOps infrastructure. He brings technical expertise across the data stack including AWS, Python, Django, Java, PostgreSQL, Hadoop, Spark, Kafka, Docker, CI/ CD, SQL, NoSQL, and more.
