Lesson 34 of 40 AI & ML Advanced 55 min

ML.NET & AI Integration

Integrate machine learning into .NET apps using ML.NET 4, Semantic Kernel for LLM orchestration, and Azure AI services.

Part 1: ML.NET AutoML

var context = new MLContext();
var data = context.Data.LoadFromTextFile<OrderData>("orders.csv");

var experiment = context.Auto().CreateRegressionExperiment(300);
var result = experiment.Execute(data, labelColumnName: "Total");

Part 2: Semantic Kernel for LLM Apps

var kernel = Kernel.CreateBuilder()
  .AddAzureOpenAIChatCompletion("gpt-4o", endpoint, apiKey)
  .Build();

var result = await kernel.InvokePromptAsync(
  "Summarize this order: {{$order}}",
  new KernelArguments { ["order"] = orderJson });

Part 3: Copilot Studio Plugin

Expose your .NET service as a Copilot Studio plugin:

// Kernel function exposed as AI plugin
[KernelFunction, Description("Gets order by ID")]
public async Task<string> GetOrder(
  [Description("The order identifier")] int orderId)
=> (await _svc.GetAsync(orderId)).ToString();

Part 4: ONNX Model Inference

Run pre-trained ONNX models (PyTorch, scikit-learn) in .NET:

var session = new InferenceSession("model.onnx");
var inputs = new NamedOnnxValue[] { NamedOnnxValue.CreateFromTensor("input", tensor) };
var outputs = session.Run(inputs);
var prediction = outputs.First().AsEnumerable<float>().ToArray();

ML.NET & AI Integration

Part 1: ML.NET AutoML

Part 2: Semantic Kernel for LLM Apps

Part 3: Copilot Studio Plugin

Part 4: ONNX Model Inference

VISUAL STUDIO 2026 MADE EASY