We Build a ML Pipeline After We Deploy

Explore the importance of building end-to-end machine learning pipelines from day one in this 32-minute EuroPython Conference talk by Alyona Galyeva. Learn why ML pipelines are necessary, when to use them, and their key building blocks for both training and inference. Discover techniques for engineering around failures, optimizing performance, debugging, and monitoring ML pipelines. Gain insights into useful open-source Python libraries that can save time in pipeline development. Ideal for data scientists, analysts, engineers, ML engineers, product owners, and Python developers working or interested in machine learning. Basic knowledge of Data Science, ML, and Python is recommended to fully benefit from this comprehensive overview of ML pipeline development and implementation.

Introduction
One stop solution
Agenda
Who am I
What is ML pipeline
Why do we need this pipeline
Why automate it
Reduce the cost of any project
When should we use it
When to scale
Building blocks
Continuous Integration
Continuous Delivery
Automated Pipeline
Continuous Delivery Process
Monitoring
Engineering
Debugging
Top 3 debugging issues
Python libraries
QA time