Multi-Tenant Machine Learning Platform Based on Kubernetes

Abstract

In this paper, we propose a flexible and scalable machine learning architecture based on Kubernetes that can support simultaneous use by huge numbers of users. Its utilization of computing resources is superior to virtual-machine-based architectures because of its container-level resource isolation and highperformance orchestration mechanism. We also describe the implementation of several important features that are designed to simplify the entire modeling lifecycle for machine learning developers. Real case studies for machine learning model development are presented that demonstrates the effectiveness of the platform in reducing the barriers to machine learning.