Azure ML Ops - 2 Wks - Operational Assessment

PROBLEM STATEMENT
Your Data Scientists have developed multiple Azure Machine Learning (ML) models and you wish to move them to production to start realizing their benefits but don't have efficient processes or have been unsuccessful in the past.

KEY CHALLENGES

• Converting Azure ML models (Models) to run at production volumes (including related ETL)
• Training the Models using production volumes
• Storing Model outputs in a usable and accessible format
• Evaluating, tracking and registering production Models
• Deploying the Model and related data transformations in a repeatable method
• Monitoring and re-training the Models once in production
• Integrating the Model into business processes

SDK MLOPS ACCELERATOR
SDK has developed a methodology with repeatable processes (leveraging SDK accelerators) to operationalize Azure Machine Learning models. The methodology identifies the key activities, milestones and resources required for each phase:

• PROTOTYPING
• TRAINING
• MODEL MANAGEMENT
• DEPLOYMENT
• OPERATIONS

WHY INVEST?

• Minimize Model Drift - Pro-active monitoring and re-training ensures models adapt to new data patterns
• Optimize Data Scientists - Repeated and automated process allow Data Scientists to focus on the science
• Realize Azure ML Benefits Sooner - Standardized processes move models through the lifecycle more efficiently
• Cost Reduction - More performant code and more efficient allocation of Azure resources

HOW THE ASSESSMENT WORKS
SDK will assess your current Azure Machine Learning environment SDK will provide a gap assessment at each phase SDK will provide a roadmap to operationalize an Azure Machine Learning model