Overview
Explore the challenges and solutions for maintaining AI applications in production through this comprehensive conference talk. Learn about Continuous Intelligence, an adaptation of Continuous Delivery practices for machine learning projects. Discover how to bridge the gap between software engineering and data science workflows, enabling faster release cycles for AI systems. Gain insights into handling code changes, model updates, data versioning, and schema modifications in ML pipelines. Examine the differences between developer and data scientist approaches, and understand how to build effective Continuous Intelligence pipelines. Through real-world examples and a demo application, acquire practical knowledge on implementing Continuous Delivery workflows for AI projects using various tools and technologies.
Syllabus
Introduction
About Thoughtworks
Open Knowledge
Continuous Delivery for ML
Autoscout
Problems
Continuous Delivery
Pipelines
Challenges
Emily Gorcenski
Code Changes
Model Changes
Data Versioning
Sampling
Schema Changes
Building Pipelines
Developers vs Data Scientists
Continuous Intelligence Pipelines
Continuous Delivery Workflow
Tools and Technologies
Demo
Example Application
Continuous Delivery Pipeline
Conclusion
Taught by
NDC Conferences