Welcome to the second two days of the Cloud AI Bootcamp. In these two days, we will focus on hands-on activities that develop proficiency in AI-oriented workflows leveraging Azure Machine Learning Workbench and Services, the Team Data Science Process, Visual Studio Team Services, and Azure Container Services. These labs assume a introductory to intermediate knowledge of these services, and if this is not the case, then you should spend the time working through the pre-requisites.
Pre-requisites can be found here. Briefly, pre-requisites include the following:
- The ability to create resources within an Azure subscription
- Familiarity with how to create resources in said subscription
Azure Machine Learning Services
- Python Proficiency and familiarity with data science workloads and techniques.
- Familiarity with Git and a Visual Studio Team Services Account to leverage collaborative features of Azure Machine Learning.
- Understand and use the Team Data Science Process (TDSP) to clearly define business goals and success criteria
- Use a code-repository system with the Azure Machine Learning Workbench using the TDSP structure
- Create an example environment
- Use the TDSP and AMLS for data acquisition and understanding
- Use the TDSP and AMLS for creating an experiment with a model and evaluation of models
- Use the TDSP and AMLS for deployment
- Use the TDSP and AMLS for project close-out and customer acceptance
- Execute Data preparation workflows and train your models on remote Data Science Virtual Machines (with or without GPUs) and HDInsight Clusters running Spark
- Manage and compare models with Azure Machine Learning
- Explore hyper-parameters on Spark using Azure Machine Learning
- Deploy and Consume a scoring service on Azure Container Service
- Collect and Analyze data from a scoring service in production to progress the data science lifecycle.
Please note: This is a rough agenda, and the schedule is subject to change pending class activities and interaction.
- Day 1
- 9-11: Introduction and Context
- 10-11: Lab 3.1: Introduction to Team Data Science Process with Azure Machine Learning
- 11-12: Lab 3.2: Comparing and Managing Models with Azure Machine Learning
- 12-1: Lunch
- 1-2:20 Lab 3.3: Behind the scenes: Docker images and Conda environments
- 2:30-3:50 Lab 3.4: Executing a data engineering or model training workflow in a remote execution environment
- 4-5: Summary and White-board Discussion
- Day 2
- 9-9:20: Review and Next Steps
- 9:20-10:20: Lab 3.5: Executing a neural network workflow remotely using GPUs
- 10:30-10:50: Introduction to Deployment and Context
- 11-12:00: Lab 4.1: Managing Models using Azure Machine Learning
- 12:00-1: Lunch
- 1:00-1:50: Lab 4.2: Deploying a scoring service to Azure Container Service (AKS)
- 2:00-2:50: Lab 4.3: Consuming the final service
- 3:00-3:50: Lab 4.4: Collect data from a scoring service
- 4:00-5:00: Q&A and Feedback for Pro AI Bootcamp
We will also use a gitter forum for discussion. Please post comments and questions here.
These materials have been tested on Windows with:
- Docker Community Edition v
- Azure Machine Learning Workbench v