Centralized Cataloging for Machine Learning Assets with Azure ML Registry (facilitated MLOps)

Registries in Azure Machine Learning are organization-wide repositories of machine learning assets such as models, environments, and components. Registries provide a central platform for cataloging and operationalizing machine learning models across various personas, teams, and environments involved in the machine learning lifecycle. Registries foster better collaboration among data science teams by offering a central platform to share and discover machine learning models and pipelines for facilitated MLOps.
☎️ Do you need any career or technical help? Book a call with me: calendly.com/m...
🔔 Subscribe for more cloud computing, data, and AI analytics videos by clicking on subscribe button so you don't miss anything.
Sample codes used in the video:
github.com/Azu...
✅ You can contact me at:
LinkedIn: / mohammad-ghodratigohar
Email: [e-mail]
Twitter: / mg_cafe01
#AzureMachineLearningRegistry # #AzureMLOps #MLCatalogue #AzureMLRegistry #MAchineLearningOperation #Azure #AI #DevOps #AzureML

[@s.sasisekhar4608]

Azure ML Wikipedia == MG TH-cam channel

[@myself4024]

🎯 Key Takeaways for quick navigation:
00:00 🎁 *Introduction to Azure Machine Learning Registry*
- Overview of using Azure Machine Learning Registry to share machine learning assets.
- Explanation of the challenges in manually moving machine learning assets between workspaces.
- Introduction to the concept of a centralized registry to facilitate asset sharing.
03:37 🧐 *Understanding Azure Machine Learning Registry*
- Definition of Azure Machine Learning Registry as a centralized place for sharing machine learning assets.
- Benefits of using Azure Machine Learning Registry, including making assets discoverable across different workspaces.
- Discussion on how Azure Machine Learning Registry facilitates MLOps processes.
07:29 🔄 *Facilitating MLOps with Azure Machine Learning Registry*
- Explanation of how Azure Machine Learning Registry supports MLOps by serving as a reliable source for machine learning assets.
- Overview of using the registry as a centralized place for grabbing and deploying machine learning assets in different environments.
- Example of how multiple environments (Dev, Test, Prod) can access assets from the registry for streamlined MLOps.
10:40 🏗️ *Creating and Using Azure Machine Learning Registry*
- Demonstration of creating a registry in Azure Machine Learning workspace.
- Overview of shared components, environments, and models within the registry.
- Explanation of the flexibility to define access permissions and role-based access control for the registry.
13:54 🚀 *Hands-On: Using Azure Machine Learning Registry*
- Walkthrough of cloning the Azure Machine Learning examples repository and navigating to the registry-related notebook.
- Explanation of authenticating to both the Azure Machine Learning workspace and the Azure Machine Learning Registry.
- Introduction to the notebook steps, including generating random numbers for asset versioning.
18:38 📝 *Running the Notebook: Registering Components and Models*
- Running code to create an environment and component in the Dev environment.
- Registering the created environment and component in the Azure Machine Learning Registry.
- Explanation of how the registry facilitates sharing assets across different workspaces.
23:13 🔄 *Continuing with Model Registration and Deployment*
- Continuing the notebook to register the training code (model) in the Azure Machine Learning Registry.
- Overview of how the model registration enables sharing the model across different workspaces.
- Discussion on deploying the model from the registry to a specific workspace for deployment.
23:52 📦 *Registering Components in Azure ML Registry*
- Registering components in Azure ML Registry using a YAML file.
- YAML file defines the schema for the component, specifying inputs, outputs, and the code.
- Components can be registered in the Azure ML Registry for organization-wide use, promoting code scalability and reusability.
27:21 🧪 *Specifying Environment for Training Code in Azure ML*
- Defining the environment needed for running the training code.
- Using a curated default environment or a custom environment registered in Azure ML Registry.
- Ensuring that the specified environment is used when registering the training code as a component.
29:37 🔄 *Retrieving and Running a Component from Azure ML Registry*
- Retrieving a registered training code component from Azure ML Registry.
- Running a training pipeline in Azure ML Dev Workspace using the retrieved code.
- Connecting to the workspace and executing the training job with specified inputs.
31:57 🚀 *Registering Model in Azure ML Workspace and Registry*
- Registering the trained model in the Azure ML Workspace.
- Copying the registered model from the workspace to the Azure ML Registry.
- Enabling sharing and accessibility of the model across the organization via the registry.
33:22 🚢 *Deploying Model from Azure ML Registry to Workspace*
- Deploying a model from Azure ML Registry to the Azure ML Workspace for online endpoint creation.
- Managing the deployment process, including specifying compute type and percentage of traffic.
- Testing the deployed model with scoring samples to ensure functionality.
Made with HARPA AI