MLOps with Flyte - Challenges and Solutions in Machine Learning Operations
Offered By: CNCF [Cloud Native Computing Foundation] via YouTube
Course Description
Overview
Explore MLOps challenges and solutions in this 25-minute CNCF conference talk. Discover how Flyte, an open-source orchestration tool, addresses key issues in machine learning model development and deployment. Learn about reproducibility, recoverability, maintainability, audibility, scalability, and compute management in ML workflows. Gain insights into Flyte's ML-specific features and best practices for fast model development and effective productionization of ML code. Dive into topics such as software engineering practices, technical debt, incremental development, memoization, collaboration, and organizational scaling. Understand Flyte's building blocks, including tasks and workflows, and explore its architecture and features like intra-task checkpointing. Acquire valuable knowledge to enhance your MLOps processes and streamline machine learning projects.
Syllabus
Intro
Software Engineering Practices
Technical Debt
Develop Incrementally and Constantly Iterate @ Scale
Memoize, Recover and Reproduce
Collaboration & Organizational Scaling
Extend Simply
Building Blocks: Tasks
Workflows
Workflow Modalities
#1: Execute and Interact
Features for Platform Folks
Architecture Overview
#1: Intra-task Checkpointing
MLOps Best Practices
Taught by
CNCF [Cloud Native Computing Foundation]
Related Courses
Data Science in Action - Building a Predictive Churn ModelSAP Learning Applied Data Science Capstone
IBM via Coursera Data Modeling and Regression Analysis in Business
University of Illinois at Urbana-Champaign via Coursera Introduction to Predictive Analytics using Python
University of Edinburgh via edX Machine Learning con Python. Nivel intermedio
Coursera Project Network via Coursera