Simple RAG Application in .NET (Clean Architecture + Aspire)

I built a small Retrieval-Augmented Generation (RAG) sample using .NET 10, just to demonstrate how easily we can wire up:

• Vector search with PostgreSQL (pgvector)
• LLM integration
• Clean Architecture
• Local orchestration using .NET Aspire

You can find the source code here.
GitHub Repo: https://github.com/rineshpk/dotnet-rag-pgvector

What this repo demonstrates

• Minimal RAG pipeline
• Embedding storage using PostgreSQL + pgvector
• Semantic search using cosine similarity
• Clean Architecture structure
• Aspire orchestration for local development

This project is orchestrated using .NET Aspire, making local setup extremely simple.

Just clone the repo and run: dotnet run --project AppHost.
Prerequisites: .Net 10, Docker

Open the aspire dashboard in browser. Ollama and PostgreSQL will take some time to pull their images the first time. Wait for all services to start.

You can see all services running and connected in one place.

The application stores embeddings directly in PostgreSQL using pgvector.

The API is exposed with a clean UI for testing endpoints.

Here’s an example of querying the system with a design pattern question and getting an LLM-powered response.

If you're exploring:

• RAG in .NET
• pgvector usage
• Aspire-based orchestration

This repo gives you a clean starting point.

👉 Check it out here