Submit a talk on data

Submit talks on data engineering, data science, machine learning, big data and analytics through the year – 2018

Machine Learning on Kubernetes using Kubeflow

Submitted by Sanket Sudake (@tripples) on Wednesday, 16 May 2018

videocam
Preview video

Technical level

Intermediate

Status

Submitted

Vote on this proposal

Login to vote

Total votes:  +2

Abstract

Deploying applications with containers is now a de-facto standard & Kubernetes is a preferred orchestrator for deploying containers. Using kubernetes to build/train/deploy machine Learning application is desired considering out-of-box feature which kubernetes provides like autoscaling, self-healing, rolling upgrade support etc.

Kubeflow is an open-source machine learning toolkit built on top of kubernetes which helps you in different stages of ML application development lifecycle. This talk provides an overview of Kubeflow and how it can be used for ML development with a sample demos.

Outline

  • Advantages of using Kubernetes for Machine Learning(ML) application
  • What is Kubeflow and purpose of using it
  • Deploying Kubeflow on Kubernetes with Ksonnet
  • Using Jupyterhub for model development
  • Using GPUs with Kubeflow
  • Training model using Kuberntes Custom Resource Definitions(CRD)
  • Distributed tensorflow on kubernetes with kubeflow
  • Serving models

Speaker bio

I work as Technical Lead at Infracloud Technologies. Recently, I started exploring different infrastrucutre related problems in Machine Learning space. I am also an active contributor to Openstack. My core interest areas are distributed systems and networking technologies. Prior to this he worked at Veritas as Linux Kernel Engineer.

Links

Slides

https://docs.google.com/presentation/d/1aLgBEfGdMrNE7I3u6NOW2oZ32V8CUI_pwSdoxAy_DD0/edit?usp=sharing

Preview video

https://youtu.be/7GwnvDn7CJk

Comments

Login with Twitter or Google to leave a comment