[MLOps] CI/CD in the age of Machine Learning

Schedule

Room: Room 203

April 5

14:00–14:25

Abstract

Machine learning models are a new artifact to build, version and deploy, explore there impacts on your architecture.

Prerequisites

Having dealt at least once with deploying a machine learning model in production.

Description

CI/CD is a well-known tool for craftsmen, enabling them to build and deploy their applications automatically and robustly. ML brings an additional artifact to building the Machine Learning model. Extracted from the book Culture MLOps, in this talk we'll explore the impact of ML model management on CI/CD. Based on 3 needs (making a prototype, starting to develop a product, scaling up), we'll see where to build the model, in which register to version it, and how to deploy a new version.

You'll leave with different patterns that can be applied in your own context.

Translated with DeepL.com (free version)