NEPCODER

Courses in AI Engineering for .NET

1. Course 1

   AI Fundamentals

   How language models actually behave, assuming no prior AI knowledge. The prerequisite for everything else in the track.

   6 lessons

   beginner6 lessons · 90 min

2. Course 2

   Prompt Engineering

   System versus user prompts, few-shot examples, structured output, evaluation and prompt versioning — treated as code, not folklore.

   • Requires: AI Fundamentals

   beginnerComing soon

3. Course 3

   Embeddings & Vector Databases

   Vector representation, similarity, chunking, indexing, and how to choose a store you will not regret.

   • Requires: AI Fundamentals

   intermediateComing soon

4. Course 4

   RAG

   Retrieval pipelines end to end: chunking strategy, re-ranking, grounding, citation, and measuring whether any of it helped.

   • Requires: Embeddings & Vector Databases
   • Suggested: Prompt Engineering

   intermediateComing soon

5. Course 5

   Semantic Kernel

   Kernels, plugins, planners and memory — Microsoft’s orchestration layer, in idiomatic .NET.

   • Requires: Prompt Engineering
   • Suggested: RAG

   intermediateComing soon

6. Course 6

   Microsoft Agent Framework

   Agent primitives, tools, state and multi-agent coordination on Microsoft’s framework.

   • Requires: Semantic Kernel

   advancedComing soon

7. Course 7

   MCP (Model Context Protocol)

   Protocol design, building servers and clients in C#, and exposing tools that other people’s agents can call safely.

   • Requires: Prompt Engineering
   • Suggested: Semantic Kernel

   advancedComing soon

8. Course 8

   AI Agents

   Patterns, tool use, planning loops, guardrails, evaluation and the failure modes nobody demos.

   • Requires: Microsoft Agent Framework
   • Suggested: RAG

   advancedComing soon

9. Course 9

   Azure AI

   Azure OpenAI and AI Search in production: deployment, quotas, cost control and monitoring.

   • Requires: AI Fundamentals
   • Suggested: RAG

   advancedComing soon

There is no account, and no server

Nothing here asks you to sign in, because there is nothing to sign in to. The whole site is prerendered static files. What you complete, answer, write and prefer is kept in this browser’s local storage and never sent anywhere — which also means clearing site data erases it. The export file on your profile is how you back it up or move it to another machine.

The one exception is worth naming: the pages you visit are counted in aggregate by Vercel Analytics, so it is possible to tell which lessons get read. Nothing you type, answer or complete is part of that.

What it deliberately does not do

• No certificates. The completion summary is your own record of what you finished, and it is not accreditation.
• No streaks, badges or leaderboards. A counter that resets because you had a sprint deadline is worse than no counter.
• No code runner. Exercises are written for the SDK you already have installed. You run them where you would run any other .NET project.
• Nothing that needs a server to adjudicate. No comments, no ratings, no per-learner analytics. Faking any of them locally would produce a feature that lies to you.

It is open, all of it

The lessons, the curriculum and the code that renders them are one public repository, together with the architecture notes and the decision records explaining why it is built this way. If a lesson is wrong, the fix is a pull request.