This course is for developers who create and run machine
learning applications on HPE Ezmeral Container Platform
5.3. The course teaches how to deploy clusters and provide
real-life prediction analysis for specific use cases. The course
consists of 30% lecture and 70% lab exercises.
COURSE OBJECTIVE:
During this course, you will learn how to:
• Set up the project repository
• Create a training cluster
• Create a Jupyter notebook and attach it to a
training cluster
• Run through an example of a typical machine
learning workflow
• Operationalize your model
• Make a prediction (inference)
• Obtain in-depth knowledge of HPE Ezmeral
Container Platform 5.3 ML Ops
• Apply best practices to help accelerate
the development of user-based prediction
analysis
TARGET AUDIENCE:
System developers, big data application
developers, business analysts, data
scientists, data engineers.
COURSE PREREQUISITES:
• AI/ML application administration
experience (Spark, Jupyter Notebook,
Tensorflow, etc.) • Experience in machine learning lifecycle
(e.g. model training/development and
model deployment) • Bash/shell/python scriptin
COURSE CONTENT:
HJ7H2S (hpe.com)
Machine Learning Ops Overview • Creating an ML Ops tenant
• External authentication
• Project repository
• Source control
• Model registr
• Training
• Deployments
• Data sources
• App store
• Notebooks HPE
Personas Overview • Platform administrator (site
administrator)
• Project administrator
• Project member
Project Repository Setup • Initial access to HPE Ezmeral
Container Platform
• Setting up ML Ops environment and project repository
• ML Ops clusters
Training Cluster Setup • Creating a training cluster
• Training cluster configurations
• Training cluster
• Spark training
• Accessing Python training cluster outside of HPE
Ezmeral Container Platform
• General notes on training clusters
Notebook Setup • Creating a notebook cluster
• Notebook cluster configuration
• More details on notebooks on ML Ops
• Create notebook with training cluster
• Review
• Training first model
Model Registry and Deployment • Model registry
• Model registry configurations
• More details on model registry
• Deployments (Method 1)
• Deployments (Method 2)
• Deployments clusters
• Register and deploy the model
Inference • “Ready” deployment cluster
• Doing inference
• Walkthrough of scoring script
• Local notebook to ML Ops training cluster
Lab 1: Initial Access to HPE Ezmeral Container
Platform • Task 1: Initial log-on to HPE Ezmeral Container
Platform
Management Console
• Task 2: Lab system setup
• Task 3: Initial log-on to controller
Lab 2: Setting Up ML Ops Environment and
Project Repository • Task 1: Set up the ML Ops environment
• Task 2: Install and register app from App Catalog
• Task 3: Setup the project repository
Lab 3: Create Training Clusters • Task 1: Create training
cluster
Lab 4: Create Notebooks with Training Cluster • Task 1:
Create notebook with training cluster
Lab 5: Training First Model • Task 1: Login to Jupyter hub •
Task 2: Training the model
Lab 6: Register and Deploy the Model • Task 1: Register the
model • Task 2: Deploy the model
Lab 7: Inference • Task 1: Generate prediction requests
Lab 8: Local Notebook to ML Ops Training Cluster • Task 1:
Making required file configurations
• Task 2: Accessing training cluster through Jupyter
Notebook
• Task 3: Training the model through local notebook
Lab 9: Spark Deployment • Task 1: Setup Spark deployment
environment
• Task 2: Stopping cluster in AIML tenant
• Task 3: Create Spark training cluster
• Task 4: Create Spark notebook cluster
• Task 5: Train the used car pricing model
• Task 6: Register new model
• Task 7: Deploy the model
• Task 8: Inference
FOLLOW ON COURSES:
Not available. Please contact.