Integrating Azure Machine Learning with .NET Applications

Overview

Azure Machine Learning (AML) provides a robust suite for building, training, and deploying ML models. This guide covers the steps required to consume those models directly from .NET using the Azure.ML SDK and REST endpoints.

Prerequisites

• .NET 6.0 or later
• Azure subscription with an AML workspace
• Visual Studio 2022 or VS Code
• Azure CLI installed

Step‑by‑Step Guide

1. Create or select a model in your AML workspace and deploy it as a real‑time endpoint.
2. Generate the endpoint key from the Azure portal.
3. Add the Azure.AI.MachineLearning SDK to your .NET project:

   dotnet add package Azure.AI.MachineLearning

4. Consume the endpoint from code:

   using Azure;
   using Azure.AI.MachineLearning;
   using System.Text.Json;
   
   var client = new MachineLearningClient(new Uri("https://<your‑workspace>.cognitiveservices.azure.com/"), new AzureKeyCredential("<endpoint‑key>"));
   
   var request = new
   {
       input = new { data = new[] { 5.1, 3.5, 1.4, 0.2 } }
   };
   
   Response<BinaryData> response = await client.PredictAsync("<endpoint‑name>", RequestContent.Create(JsonSerializer.Serialize(request)));
   Console.WriteLine(response.Content.ToString());

5. Handle the response and integrate the prediction into your application logic.

Full Sample Project

Clone the sample repository for a ready‑to‑run project:

git clone https://github.com/msdn-community/azure-ml-dotnet-sample.git