1 of 4

Machine Learning on Kubernetes using Kubeflow

Anthill Inside 2018

2 of 4

Sanket Sudake

  • Technical Lead, Infracloud Technologies
  • Cofounder, QuodeIt
  • Ex-Veritas, Kernel developer
  • ML enthusiast, Kubernetes

​

3 of 4

Agenda

  • Why Kubernetes for ML application
  • Kubeflow and Purpose
  • Deploying Kubeflow
  • Using Jupyterhub for model development
  • Using GPUs with Kubeflow
  • Training model using Kubernetes CRDs
  • Distributed tensorflow with kubeflow
  • Serving models

4 of 4

Key Takeaways

  • Using containerised solutions across Build/Train/Deploy lifecycle of ML application
  • How to make your ML application resilient, portable and scalable
  • How to deploy ML applications using declarative approach
  • Using GPUs/CPUS efficiently via advanced scheduling capabilities of Kubernetes