Overview
Explore the challenges and solutions for maintaining AI applications in production through Continuous Intelligence practices. Learn how to adapt Continuous Delivery principles from software engineering to the data science world, enabling faster release cycles and more efficient maintenance of machine learning models. Discover the differences between data scientist and software engineer workflows, and how to bridge these gaps. Dive into the concept of Continuous Intelligence, including build pipelines, delivery pipelines, and technology stacks. Examine a practical demo of a Continuous Intelligence workflow, covering aspects such as model packaging, data versioning, evaluation stages, and continuous monitoring. Gain insights on bringing data scientists and developers closer together, handling supervised learning scenarios, and implementing effective testing strategies for AI applications.
Syllabus
Introduction
Continuous Delivery for Machine Learning
Build Pipeline
Challenges
Reasons
More types of change
Pipelines
Stages
Technology stack
Demo
Delivery Pipelines
Continuous Intelligence Workshop
How it works
Red pipeline
DBC pull
Redeploy
Continuous Intelligent Cycle
Questions
How to package a model
Bringing data scientists and developers closer
Data versioning
Evaluation stage
Test stage
Supervised learning
Working together
Continuous monitoring
Monitoring
Taught by
NDC Conferences