The Kubeflow project makes deployments of machine learning (ML) workflows on Kubernetes simple, portable and scalable. Kubeflow comes with Jupyter notebook, training and inference using Tensorflow, hyperparameter tuning using Katib, end-to-end automated deployment pipelines using Argo, hyperparameter tuning using Katib, and much more.
This talk will explain why and how Kubernetes is well suited for single- and multi-node distributed training, training your models, and deploying your models for inference in production.
Specifically it will show how to use KubeFlow and TensorFlow for your machine learning needs. We will also show to setup machine learning pipelines and set up visualization tools like TensorBoard for monitoring.
We will also discuss distributed training using Horovod.
Arun Gupta is a Principal Technologist at Amazon Web Services. He is responsible for the Cloud Native Computing Foundation (CNCF) strategy within AWS, and participates at CNCF Board and technical meetings actively. He works with different teams at Amazon to help define their open source strategy. He has built and led developer communities for several years. He has extensive speaking experience in 45+ countries on myriad topics. Gupta also founded the Devoxx4Kids chapter in the US and continues to promote technology education among children. A prolific writer, author of several books, an avid runner, a globe trotter, a Docker Captain, a Java Champion, a JUG leader, he is easily accessible at @arungupta on twitter.
|Talks by tracks||Talks by session types||List of Speakers||Schedule|