Overview
Explore the intricacies of productionalizing machine learning models through CI/CD design using MLflow in this 36-minute talk from Databricks. Learn how to automate the complex pipeline and feedback loop of model deployment and integration. Discover how MLflow and the model registry can simplify building a robust CI/CD pattern for any given model. Walk through an end-to-end example of designing a CI/CD process for model deployment, implementing it with MLflow and automation tools. Gain insights into traditional CI/CD, machine learning CI/CD, and the role of MLflow in streamlining these processes. Understand the control model flow, model registry, deployment strategies, and the execution of build and deploy clusters. By the end of this talk, acquire the knowledge to effectively integrate MLflow with continuous integration, continuous development, and continuous deployment tools for efficient model productionalization.
Syllabus
Intro
Introductions
Agenda
CICD Definition
Traditional CICD
Machine Learning CICD
MLflow Example
Control Model Flow
Model Registry
Deployment
Deploy
Build Cluster
Execute Cluster
Summary
Taught by
Databricks