Machine Learning Made Easy on Kubernetes. DevOps for Data Scientists
summary
A Game of Thrones image classifier is built from the TensorFlow “retrain” Inception sample, trained on Kubernetes as a TFJob, stored in Azure Storage, and served via TensorFlow Serving behind a Kubernetes Service; hyperparameter tuning runs nine Helm-driven combinations and can scale to many more with Katib; demos tie Kubeflow pipelines, Pachyderm for data, Argo-style automation, and Azure virtual nodes to run pay-per-use container instances. TFJob + TensorFlow Serving 9 hyperparameter runs
AI-written summary · about
chapters (30)
- 0:00Introduction
- 0:57Demo
- 3:01Machine Learning
- 5:55Kubernetes
- 10:33MLOps
- 11:46Pipelines
- 13:14Demonstration
- 13:55Code
- 14:39Docker Container
- 15:03Tensorflow
- 16:34Deploying TensorFlow
- 16:50Do I want my data scientists
- 17:36Cubeflow dashboard
- 18:15Azure storage
- 19:25Service
- 20:21Service IP
- 20:39Curl
- 21:02Web App
- 21:18Game of Thrones
- 21:50Hyperparameters
- 24:13Virtual Nodes
- 25:13Container Instances
- 25:41Install Helm
- 26:23Tensor Board
- 26:39Kubernetes Instances
- 27:10Khateeb
- 29:13Kubernetes Pipelines
- 31:03Training Logs
- 32:28Notebooks
- 34:17Demo QA
related talks
37mA Day in the Life of a Data Scientist. Conquer ML Lifecycle on Kubernetes
46mDeploying scalable Machine Learning models with Kubernetes
42mKubernetes for Machine Learning: Managing and Scaling AI Workloads
22mThe Containerization of Machine Learning: TensorFlow: Kubernetes, and Kubeflow
31mKubernetes for Machine Learning: Managing and Scaling AI Workloads
26mMachine Learning Meets K8s
46mAccelerating Machine Learning DevOps with Kubeflow
46mMachine Learning with Kubernetes