Generative AI for .NET Developers with Microsoft Foundry

Descripción

This course is being rebuilt for the Microsoft Foundry era. It is currently being modernized to align with Microsoft Foundry, the native Microsoft AI Extensions library, Azure AI Search, RAG, agents, evaluation, security, observability, and production deployment. New lectures are being added while older Azure AI Studio, Azure ML, and legacy Azure AI Services content is being reviewed, replaced, or moved into optional refresher sections. Generative AI development on Azure has changed significantly. Modern .NET developers now need more than basic prompt demos or simple Azure OpenAI calls. Professional AI applications require secure model access, structured outputs, Retrieval-Augmented Generation, vector search, tool calling, agents, evaluation, observability, and production deployment practices. This course teaches .NET developers how to build production-minded generative AI applications using Microsoft Foundry, Azure OpenAI-compatible models, Azure AI Search, Microsoft AI Extensions, RAG, agents, security, evaluation, and deployment. The main project is VibeCast, an enterprise-style multimodal AI application. Through the project, you will build practical workflows for content generation, structured summaries, image and audio processing, document grounding, tool calling, agentic orchestration, security controls, monitoring, and deployment. This is not a generic AI theory course and not a pure certification cram course. It is a practical engineering course for developers who want to build real AI-enabled applications with C# and .NET. You will learn how to: • Build modern .NET AI applications using Microsoft Foundry and Azure OpenAI-compatible model deployments. • Use IChatClient, and IEmbeddingGenerator to keep your code testable and provider-flexible. • Design prompt assets, structured outputs, C# records, validation flows, and bounded response repair. • Build professional RAG workflows with Azure AI Search, embeddings, vector indexes, hybrid search, semantic ranking, citations, and grounded response contracts. • Understand classic RAG, agentic retrieval, and Foundry IQ-style knowledge grounding. • Build typed tools and agent workflows with safe execution boundaries. • Secure AI applications against prompt injection, indirect prompt injection, data leakage, and unsafe tool execution. • Add evaluation datasets, red-team tests, OpenTelemetry traces, token/cost monitoring, and production dashboards. • Containerize and deploy .NET AI applications using Azure-ready architecture patterns.

Lo que aprenderás

• Build production-ready generative AI applications with .NET 10, C#, ASP.NET Core, and Microsoft Foundry.
• Use with IChatClient and IEmbeddingGenerator to create provider-flexible AI application services.
• Connect .NET applications to Azure OpenAI-compatible model deployments using secure configuration and Azure identity patterns.
• Design prompt assets as versioned application resources instead of hardcoded strings.
• Generate schema-constrained structured outputs and validate them with C# records and domain rules.
• Build Retrieval-Augmented Generation workflows using Azure AI Search, embeddings, vector indexes, hybrid search, semantic ranking, and citations.
• Understand classic RAG, agentic retrieval, and Foundry IQ-style knowledge grounding patterns.
• Build tool-using AI workflows with typed C# functions, authorization checks, approval gates, and safe execution boundaries.
• Understand when to use deterministic C# code, model calls, workflows, or AI agents.
• Build multimodal workflows for text, image, vision, speech, audio, and content extraction scenarios.
• Secure generative AI applications against prompt injection, data leakage, unsafe tool calls, and malicious documents.
• Evaluate, monitor, trace, containerize, and deploy .NET AI applications using OpenTelemetry, Application Insights, GitHub Actions, and Azure hosting options.

Requisitos

• Basic C# knowledge.
• Basic ASP.NET Core or Web API familiarity.
• Basic Azure familiarity is helpful but not required.
• No machine learning or data science background required.
• An Azure subscription is recommended for full hands-on practice.
• Visual Studio, Visual Studio Code, or another .NET-capable IDE.