{"enrichment":{"faq":[{"a":"azure-ml-foundry-workspace guides workspace creation through the Azure portal, az ml CLI v2, PowerShell Az.MachineLearningServices, or Terraform. Start by defining resource group, region, and storage account. Use `az ml workspace create` with `--name`, `--resource-group`, and `--location` flags. For AI Foundry projects, create a hub first, then projects within it. Enable managed networking with `--public-network-access Disabled` and configure private endpoints for secure connectivity to dependent resources like ACR and storage.","q":"How to create Azure ML workspace?"},{"a":"azure-ml-foundry-workspace covers managed online endpoints as one of three deployment options alongside batch and serverless endpoints. Managed online endpoints provide real-time inference with autoscaling, traffic splitting, and A/B testing capabilities. Deploy models via `az ml online-endpoint create` specifying compute SKU, instance count, and traffic rules. Monitor endpoint health through Azure portal or CLI. Managed endpoints handle infrastructure provisioning automatically, supporting both standard and GPU-backed compute for low-latency predictions.","q":"What are managed online endpoints in Azure ML?"},{"a":"azure-ml-foundry-workspace explains private endpoint configuration for secure workspace networking. Enable managed networking during workspace creation or post-deployment. Create private endpoints for workspace, blob storage, file share, key vault, and container registry. Configure DNS zones to resolve private IPs. Use `--public-network-access Disabled` in CLI commands. Managed private endpoints simplify setup by auto-provisioning connections to dependent Azure services. Verify connectivity via diagnostic tools and check network security group rules blocking traffic.","q":"How do I set up private endpoints for Azure ML workspace?"},{"a":"azure-ml-foundry-workspace provides debugging workflows for deployment failures. Stream job logs with `az ml job stream --name <job-id>`. Check endpoint logs via `az ml online-endpoint get-logs` or batch job logs with `az ml batch-endpoint get-logs`. Review managed identity RBAC permissions\u2014ensure the workspace identity has Contributor or custom roles on compute and storage resources. Inspect container registry authentication, networking rules, and compute cluster status. Enable diagnostic settings to send logs to Log Analytics for deeper troubleshooting of provisioning and runtime errors.","q":"How to troubleshoot Azure ML deployment errors?"},{"a":"azure-ml-foundry-workspace references GPU SKU selection for compute clusters and online endpoints, including H100, A100, V100, and A10 options. SKU availability varies by region and subscription quota. Specify GPU SKU via `--vm-size` parameter in cluster creation: `az ml compute create --vm-size Standard_NC24ads_A100_v4`. For managed endpoints, select compute SKU matching workload latency and throughput needs. Monitor GPU utilization and cost. Request quota increases through Azure portal if SKU unavailable in target region.","q":"What GPU SKUs does azure-ml-foundry-workspace support?"},{"a":"azure-ml-foundry-workspace includes Terraform automation for workspace provisioning. Define `azurerm_machine_learning_workspace` resource with storage account, key vault, container registry, and application insights dependencies. Use `azurerm_private_endpoint` for secure networking. Configure managed identity via `identity` block with `type = \"SystemAssigned\"`. Reference `azurerm_machine_learning_compute` for clusters and instances. Store Terraform state securely. Combine with `az ml` CLI for post-deployment configuration like endpoint creation and RBAC role assignments.","q":"How to automate workspace provisioning with Terraform?"}],"shadow_tags":["infrastructure-as-code","model-serving","network-isolation","gpu-acceleration","identity-management","container-orchestration","real-time-inference","batch-processing","devops-automation","troubleshooting-diagnostics"],"summary_rewrite":"Complete reference for provisioning and configuring Azure Machine Learning workspaces and AI Foundry projects, covering workspace hierarchy, managed networking with private endpoints, compute targets, and endpoint deployment. Includes az ml CLI v2 and PowerShell Az.MachineLearningServices commands, managed identity setup, ACR integration, and diagnostic workflows for troubleshooting compute and deployment logs."},"files":[{"bytes":17085,"path":"plugins/azure-master/skills/azure-ml-foundry-workspace/SKILL.md","sha256":"d89e7f02f88df7cf015ba7f4180a1dd088e14778fc8388884decee6549bef84a","url":"https://skillfed.io/files/JosiahSiegel/claude-plugin-marketplace/azure-ml-foundry-workspace/9495ebdd/SKILL.md"}],"id":"JosiahSiegel/claude-plugin-marketplace/azure-ml-foundry-workspace","links":{"html":"https://skillfed.io/JosiahSiegel/claude-plugin-marketplace/azure-ml-foundry-workspace","md":"https://skillfed.io/JosiahSiegel/claude-plugin-marketplace/azure-ml-foundry-workspace.md","repo":"https://github.com/JosiahSiegel/claude-plugin-marketplace"},"meta":{"agents_supported":[],"first_seen":"2026-07-28","forks":10,"language":"Shell","last_updated":"2026-06-18","license":"MIT","name":"azure-ml-foundry-workspace","publisher":"JosiahSiegel","stars":49},"relations":{"similar":[{"id":"JosiahSiegel/claude-plugin-marketplace/ml-cloud-deployment"},{"id":"BagelHole/DevOps-Security-Agent-Skills/azure-aks"},{"id":"JosiahSiegel/claude-plugin-marketplace/container-apps-gpu-2025"},{"id":"JosiahSiegel/claude-plugin-marketplace/azure-openai-2025"},{"id":"JosiahSiegel/claude-plugin-marketplace/deployment-stacks-2025"},{"id":"JosiahSiegel/claude-plugin-marketplace/aks-automatic-2025"},{"id":"BagelHole/DevOps-Security-Agent-Skills/azure-sql"},{"id":"pulumi/agent-skills/pulumi-arm-to-pulumi"},{"id":"rysweet/amplihack/azure-admin"},{"id":"BagelHole/DevOps-Security-Agent-Skills/azure-networking"}]},"slug":{"owner":"JosiahSiegel","repo":"claude-plugin-marketplace","skill":"azure-ml-foundry-workspace"},"version":"9495ebdd"}