Azure ML - Deploying ML models using Containers and ACI
We are starting a Series on Cloud Native AI that covers hands-on training on AI/ML Development, Training and Deployment on different Public and Hybrid clouds and Cloud-Native Technologies.
Within the Cloud Native, Azure AI/ML is a sub-series. This video is a complete hands-on guide to training ML models and deploying them using docker containers as Azure Container Instances (ACI).
This video starts with a walkthrough of:
Azure ML WORKSPACE
Azure ML EXPERIMENTS
Azure experiment RUNS
Azure ML MODELS
Azure ML DEPLOYMENTS
The video then covers hands-on walkthrough of training a sample model (diabetes prediction), followed by:
Uploading the model pkl to azure ml workspace
Registering the uploaded model as a MODEL instance in the workspace
Version control of ML MODELS within Azure ML Workspace
Automatically creating Scoring scripts and preparing the container deployment folder
Describing a python environment programmatically and then serializing it to the container deployment folder.
Deploying a specific model version as a docker container using Azure ACI
Invoking model inference/ scoring of the deployed model in online and batch modes using:
Azure ML_SDK
Python request library
Automatically deleting the ML model deployment across any type of deployment.