MLOps on Azure Machine Learning: 1-Day Workshop

MLOps on Azure Machine Learning workshop is designed to provide the knowledge and share real project experience on how to effectively manage and optimize the operations of ML software using Azure Machine Learning. This workshop aims to help Data Science organizations in Azure ML adoption, as well as implementation of proper MLops practices and standards. Content of the workshop will be delivered by Senior AI Engineers with technical leadership experience in complex productization projects for large organizations. The workshop will begin with an introduction to key MLOps challenges and how they are addressed by AzureML platform, such as organization of experimentation environment, code, data and model lifecycle management, metrics tracking, production migration, monitoring, debugging, and security. The second part of the workshop is focused on Azure ML functionalities and architecture review focusing on topics like managing data, jobs, components, pipelines, environments, models, and endpoints. The workshop will conclude with walk through end-to-end example project scenario and open discussion for further questions. The agenda presented below can be tailored to meet specific needs of given organization.

Lingaro can also help in further steps of AzureML and MLOps adoption providing professional services for:

• AzureML setup and configuration
• ML solution productization and scalling
• MLOps best practices implementation

Don't miss out on this chance to gain the skills and knowledge you need to improve ML adoption and MLOps practices on Azure Machine Learning platform.

Agenda:

1. Intro

• ML Software, with underlying ML model, transforms continuously changing input data to the highest value output data.
• MLOps - operations, practices, and processes for ML Software

2. MLOps challenges

• Experimentation Environment
• Code Management
• Data Management
• Model lifecycle
• Metrics / KPIs tracking
• Code, Data and Model CI/CD and Environments
• Production migration
• Monitoring
• Production Data issues
• Production Infrastructure issues
• Debugging
• Processes and data lineage
• Data / model rollbacks
• Security
• Organization standards
• ML Project roles and efficiency

3. Azure ML

• Platform Architecture
• Azure ML Assets / Objects i. Data ii. Jobs iii. Components iv. Pipelines v. Environments vi. Models vii. Endpoints
• Managed Compute
• Orchestration
• Linked Services
• Data Labelling
• Python SDK v1 vs. V2
• Shared Assets
• Solutions to 2.a. – 2.p. challenges

4. Examples

• Use Case
• Solution Design
• ML Pipelines

5. Discussion & deep dive