Production AI Engineering

What You'll Learn

• Implement resilience patterns for AI service calls
• Secure AI applications against prompt injection and data leakage
• Set up observability with OpenTelemetry for AI workloads
• Optimize costs across Azure OpenAI deployments
• Design scalable AI architectures for production traffic

Prerequisites

• C# and .NET 8+
• Experience building AI features (see Semantic Kernel or Agents paths)
• Azure subscription

Learning Path Articles

1. 1

   Securing .NET AI Apps: Key Vault, PII Filtering & Prompt Injection Defense

   Production security for .NET AI apps: Key Vault for secrets, PII redaction middleware, content filtering, and prompt injection defense patterns with code.

   Intermediate 12 min read university
2. 2

   AI Workflow Patterns in .NET: Chaining, Fan-Out, Human-in-the-Loop & Agents

   Design reliable AI workflows in .NET: prompt chaining, parallel fan-out, human-in-the-loop, and agent orchestration. Pattern code + when to use each one.

   Intermediate 13 min read university
3. 3

   OpenTelemetry for AI Apps in .NET: Trace Semantic Kernel + Track Token Costs

   Add observability to .NET AI apps: trace Semantic Kernel calls, track token costs per request, and export to Aspire Dashboard and Application Insights.

   Intermediate 14 min read university
4. 4

   Cut Azure AI Costs in C#: Token Budgets, Model Routing & Semantic Cache

   Reduce Azure AI Foundry spend in .NET 10: GPT-5.4 tier routing, semantic caching, token budgets, and Ollama for dev. April 2026 pricing examples included.

   Intermediate 15 min read university
5. 5

   Azure.AI.OpenAI v1 → v2 Migration in C#: Every Breaking Change + Fix Table

   Migrate Azure.AI.OpenAI v1 to v2: ChatRequestMessage → ChatMessage, streaming API changes, OpenAI package dependency. Compile-error fix table with code diffs.

   Intermediate 13 min read university