{"id":48147348,"url":"https://github.com/rineshpk/dotnet-rag-pgvector","last_synced_at":"2026-04-04T17:01:34.781Z","repository":{"id":346102756,"uuid":"1187958371","full_name":"rineshpk/dotnet-rag-pgvector","owner":"rineshpk","description":"This project demonstrates how to build a semantic search + RAG pipeline using .NET","archived":false,"fork":false,"pushed_at":"2026-03-22T12:20:32.000Z","size":26,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-03-22T22:33:44.493Z","etag":null,"topics":["ai","aspire-dotnet","cosine-similarity","dotnet","llm","ollama","openai","pgvector","postgresql","rag","rag-pipeline","retrieval-augmented-generation","similarity-search","vector-database","vector-search"],"latest_commit_sha":null,"homepage":"","language":"C#","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/rineshpk.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-03-21T12:28:12.000Z","updated_at":"2026-03-22T12:20:36.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/rineshpk/dotnet-rag-pgvector","commit_stats":null,"previous_names":["rineshpk/dotnet-rag-pgvector"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/rineshpk/dotnet-rag-pgvector","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rineshpk%2Fdotnet-rag-pgvector","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rineshpk%2Fdotnet-rag-pgvector/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rineshpk%2Fdotnet-rag-pgvector/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rineshpk%2Fdotnet-rag-pgvector/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/rineshpk","download_url":"https://codeload.github.com/rineshpk/dotnet-rag-pgvector/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rineshpk%2Fdotnet-rag-pgvector/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":31407387,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-04T10:20:44.708Z","status":"ssl_error","status_checked_at":"2026-04-04T10:20:06.846Z","response_time":60,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai","aspire-dotnet","cosine-similarity","dotnet","llm","ollama","openai","pgvector","postgresql","rag","rag-pipeline","retrieval-augmented-generation","similarity-search","vector-database","vector-search"],"created_at":"2026-04-04T17:01:32.288Z","updated_at":"2026-04-04T17:01:34.193Z","avatar_url":"https://github.com/rineshpk.png","language":"C#","funding_links":[],"categories":[],"sub_categories":[],"readme":"# .NET RAG with PostgreSQL pgvector + Aspire\n\nA production-style **Retrieval-Augmented Generation (RAG)** implementation in .NET using PostgreSQL `pgvector`, .NET Aspire, and pluggable LLM providers (Ollama / OpenAI).\n\n---\n\n## Overview\n\nThis project demonstrates how to build a **semantic search + RAG pipeline** using modern .NET practices:\n\n* Vector similarity search with **pgvector**\n* LLM-based answer generation (Ollama / OpenAI)\n* Clean Architecture (Domain, Application, Infrastructure)\n* **.NET Aspire orchestration** (Postgres + Ollama + API)\n* Provider-agnostic design (LLM + vector store)\n\n---\n\n## Architecture\n\n```text\nUser Query\n   ↓\nEmbedding Service (Ollama / OpenAI)\n   ↓\nVector Store (pgvector)\n   ↓\nTop-K Retrieved Documents\n   ↓\nLLM (RAG Prompt)\n   ↓\nGenerated Answer\n```\n\n---\n\n## Solution Structure\n\n```text\nDotNetRagPgvector/\n│\n├── AppHost/                 # Aspire orchestration\n├── Api/                     # Minimal API (entry point)\n├── Application/             # Use cases + abstractions\n│   ├── Abstractions/        # IEmbeddingService, IVectorStore, ILLMService\n│   ├── Services/            # RagService, IngestionService\n│   └── DTOs/                # UserInput\n│\n├── Domain/                  # Core models\n│   └── Models/\n│       ├── Document\n│       └── RetrievedDocument\n│\n├── Infrastructure/          # Implementations\n│   ├── Entities/            # EF Core Entities - DesignPatterns\n│   ├── Persistence/         # EF Core + pgvector\n│   ├── VectorStores/        # PgVectorStore\n│   ├── Embeddings/          # Ollama / OpenAI\n│   └── LLM/                 # Ollama / OpenAI\n```\n\n---\n\n## Features\n\n* ✅ Semantic search using embeddings\n* ✅ Retrieval-Augmented Generation (RAG)\n* ✅ pgvector integration with PostgreSQL\n* ✅ Clean architecture with strict separation of concerns\n* ✅ Pluggable LLM providers (Ollama / OpenAI)\n* ✅ Aspire-based container orchestration\n* ✅ Scalar UI for API testing\n\n---\n\n## Tech Stack\n\n* .NET 10\n* ASP.NET Core Minimal APIs\n* Entity Framework Core\n* PostgreSQL + pgvector\n* .NET Aspire\n* Ollama (local LLM)\n* OpenAI (optional)\n\n---\n\n## Getting Started\n\n### 1. Prerequisites\n\n* .NET 10 SDK\n* Docker Desktop (Or Podman - set Aspire container runtime to podman)\n* .NET Aspire workload\n\n```bash\ndotnet workload install aspire\n```\n\n---\n\n### 2. Run the Application\n\n```bash\ndotnet run --project AppHost\n```\n\nThis starts:\n\n* PostgreSQL (pgvector)\n* Ollama\n* API\n\n---\n\n### 3. Open API UI (Scalar)\n\nNavigate to:\n\n```text\nhttp://localhost:\u003capi-port\u003e/scalar\n```\n\n---\n\n## Testing the API\n\n### POST `/ask`\n\n```json\n{\n  \"query\": \"How to handle distributed transactions?\"\n}\n```\n\n### Example Questions\n\n* How to prevent cascading failures?\n* How to scale read-heavy systems?\n* How do microservices communicate asynchronously?\n\n---\n\n## Configuration\n\n### appsettings.json\n\n```json\n{\n  \"AI\": {\n    \"Provider\": \"Ollama\",\n    \"Ollama\": {\n      \"BaseUrl\": \"http://localhost:11434\",\n      \"EmbeddingModel\": \"nomic-embed-text\",\n      \"ChatModel\": \"phi4-mini\"\n    },\n    \"OpenAI\": {\n      \"ApiKey\": \"\",\n      \"EmbeddingModel\": \"text-embedding-3-small\",\n      \"ChatModel\": \"gpt-4o-mini\"\n    }\n  }\n}\n```\n\n---\n\n### Provider Switching\n\nSwitch between Ollama and OpenAI:\n\n```json\n\"Provider\": \"Ollama\"\n// or\n\"Provider\": \"OpenAI\"\n```\n\nNo code changes required.\n\n---\n\n## 🗄️ Database \u0026 Vector Search\n\n* Uses PostgreSQL with `pgvector`\n* Embeddings stored as `vector` column\n* Cosine similarity used for retrieval:\n\n```c#\nvar results = await dbContext.DesignPatterns\n    .OrderBy(x =\u003e x.Embedding.CosineDistance(questionVector))\n    .Take(2)\n    .ToListAsync();\n```\n\n```sql\nORDER BY \"Embedding\" \u003c=\u003e @queryVector\nLIMIT 2\n```\n\n---\n\n## Data Seeding\n\n* Automatic on startup\n* Uses real-world **architecture patterns dataset**\n* Embeddings generated via selected provider\n\n---\n\n## RAG Flow (Code-Level)\n\n```csharp\n// 1. Embed query\nvar queryEmbedding = await _embedding.GenerateAsync(query);\n\n// 2. Retrieve relevant documents\nvar docs = await _vectorStore.SearchAsync(queryEmbedding);\n\n// 3. Build prompt\n// 4. Generate response using LLM\n```\n\n---\n\n## Extensibility\n\nYou can easily extend:\n\n### Vector Stores\n\n* pgvector ✅\n* Pinecone (future)\n* FAISS (future)\n\n### LLM Providers\n\n* Ollama ✅\n* OpenAI ✅\n* Azure OpenAI (easy to add)\n\n---\n\n## Design Principles\n\n* Domain is persistence-agnostic\n* Infrastructure handles external dependencies\n* Application orchestrates the RAG pipeline\n* API acts as composition root\n\n---\n\n## Future Improvements\n\n* Hybrid search (BM25 + vector)\n* HNSW indexing for pgvector\n* Streaming LLM responses\n* Evaluation \u0026 benchmarking\n* Metadata filtering\n\n---\n\n## Contributing\n\nContributions are welcome! Feel free to open issues or PRs.\n\n---\n\n## License\n\nMIT License\n\n---\n\n## ⭐ Why This Project?\n\nThis repository demonstrates a **production-ready, provider-agnostic RAG architecture in .NET**.\n\n---\n\n## Acknowledgements\n\n* pgvector\n* .NET Aspire\n* Ollama\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Frineshpk%2Fdotnet-rag-pgvector","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Frineshpk%2Fdotnet-rag-pgvector","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Frineshpk%2Fdotnet-rag-pgvector/lists"}