Categorizing Docker Hub’s Public Images: End-to-End Machine Learning Pipeline with Docker Enterprise | Docker


A recent initiative at Docker was to rapidly develop and deploy end-to-end production-ready pipelines to enable discovery for Docker Hub. To reduce complexity and emphasize portability, the team implemented Docker Enterprise and Kubeflow to automatically classifying container images.

Docker Data Engineers will walk you through the project that leverages various data sources and machine learning to derive categories and labels for Docker images. Kubeflow facilitates the machine learning development workflow while Docker Enteprise provides the core production-ready platform with security scanning, monitoring and flexibility of deployment environments.

This session demonstrates the benefits of running machine learning pipelines on Docker Enterprise. It provides inspiration and concrete examples for engineers and data scientists considering containers to improve existing applications.

Roberto Hashioka – Software Engineer, Docker
Amn Rahman – Data Engineer, Docker