Enarro - Production-Grade RAG Application

A production-ready Retrieval-Augmented Generation (RAG) application built with .NET 10, featuring advanced document processing, streaming chat responses, and comprehensive infrastructure orchestration using .NET Aspire.

🚀 Features

Core Capabilities

• 📄 Multi-Document Processing - Batch upload and processing with parallel execution
• 💬 Context-Aware Chat - Conversation history with session management
• 🔄 Streaming Responses - Real-time chat responses using Server-Sent Events (SSE)
• 📚 Document Management - Full CRUD operations with metadata persistence
• 🔍 Semantic Search - Vector-based document retrieval with relevance scoring
• 📊 Source Citations - Automatic citation extraction with relevance tracking

Production Features

• 🗄️ PostgreSQL Integration - Persistent document metadata with EF Core
• ⚡ Redis Distributed Caching - Session management for horizontal scaling
• 📝 Structured Logging - Serilog with console and file outputs
• 🏥 Health Checks - Comprehensive monitoring for all dependencies
• 🐳 Container Orchestration - Full .NET Aspire infrastructure management

🏗️ Architecture

Technology Stack

Component	Technology
RAG Orchestration	Microsoft Kernel Memory
LLM & Embeddings	Ollama (phi3, nomic-embed-text)
Vector Database	Qdrant
Relational Database	PostgreSQL
Distributed Cache	Redis
Service Orchestration	.NET Aspire
API Framework	ASP.NET Core 10.0
Logging	Serilog

System Architecture

graph TB
    Client[Client] --> API[Enarro API]
    API --> PG[(PostgreSQL)]
    API --> Redis[(Redis Cache)]
    API --> KM[Kernel Memory]
    KM --> Qdrant[(Qdrant Vector DB)]
    KM --> Ollama[Ollama LLM]
    
    API --> HC[Health Checks]
    HC -.-> PG
    HC -.-> Redis
    HC -.-> Qdrant
    HC -.-> Ollama

Loading

📋 Prerequisites

• .NET 10 SDK
• Docker Desktop
• Ollama (for local LLM)

🚀 Quick Start

1. Clone the Repository

git clone https://github.com/yourusername/enarro.git
cd enarro

2. Pull Required Ollama Models

ollama pull phi3
ollama pull nomic-embed-text

3. Run with .NET Aspire

dotnet run --project Enarro.AppHost

This single command will:

• Start PostgreSQL container
• Start Redis container
• Start Qdrant container
• Start Ollama container
• Run database migrations
• Launch the API
• Open the Aspire dashboard

4. Access the Application

• API: https://localhost:7001
• Swagger UI: https://localhost:7001/swagger
• Aspire Dashboard: https://localhost:15888
• Health Checks: https://localhost:7001/health

📖 API Documentation

Document Management

Upload Single Document

POST /api/v1/ingest
Content-Type: multipart/form-data

file: <file>
tags: {"category": "technical", "department": "engineering"}

Upload Multiple Documents

POST /api/v1/ingest/batch
Content-Type: multipart/form-data

files: <file1>, <file2>, <file3>

List Documents

GET /api/v1/documents?page=1&pageSize=20&tag=technical

Get Document Details

GET /api/v1/documents/{id}

Delete Document

DELETE /api/v1/documents/{id}

Chat Operations

Standard Chat

POST /api/v1/chat
Content-Type: application/json

{
  "message": "What is machine learning?",
  "sessionId": "optional-session-id",
  "userId": "optional-user-id",
  "minRelevance": 0.3,
  "filters": {
    "category": "technical"
  }
}

Streaming Chat

POST /api/v1/chat/stream
Content-Type: application/json

{
  "message": "Explain neural networks",
  "sessionId": "session-123"
}

Response format (Server-Sent Events):

data: Neural
data:  networks
data:  are...
data: [DONE]

🔧 Configuration

appsettings.json

{
  "RAGConfigs": {
    "IndexName": "rag-test",
    "ChatModel": "phi3",
    "EmbeddingModel": "nomic-embed-text",
    "Retrieval": {
      "MinRelevance": 0.3,
      "MaxResults": 5,
      "AnswerTokens": 4096
    },
    "Conversation": {
      "MaxHistoryMessages": 10,
      "SessionTimeoutMinutes": 60
    },
    "DocumentProcessing": {
      "MaxConcurrentUploads": 5,
      "MaxFileSizeMB": 50,
      "AllowedExtensions": [".pdf", ".docx", ".txt", ".md", ".xlsx", ".pptx", ".json"]
    }
  }
}

🗄️ Database Schema

Documents Table

CREATE TABLE documents (
    id UUID PRIMARY KEY,
    file_name VARCHAR(500) NOT NULL,
    content_type VARCHAR(100) NOT NULL,
    size_bytes BIGINT NOT NULL,
    uploaded_at TIMESTAMP NOT NULL,
    uploaded_by VARCHAR(100),
    status VARCHAR(50) NOT NULL,
    chunk_count INTEGER DEFAULT 0,
    error_message TEXT,
    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

Document Tags Table

CREATE TABLE document_tags (
    id SERIAL PRIMARY KEY,
    document_id UUID REFERENCES documents(id) ON DELETE CASCADE,
    tag_key VARCHAR(100) NOT NULL,
    tag_value VARCHAR(500) NOT NULL,
    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

🏥 Health Monitoring

Access the health endpoint to check all service statuses:

GET /health

Response:

{
  "status": "Healthy",
  "timestamp": "2026-02-07T06:00:00Z",
  "checks": [
    {
      "name": "postgresql",
      "status": "Healthy",
      "duration": 12.5,
      "tags": ["db", "sql"]
    },
    {
      "name": "redis",
      "status": "Healthy",
      "duration": 8.3,
      "tags": ["cache"]
    },
    {
      "name": "ollama",
      "status": "Healthy",
      "duration": 45.2,
      "tags": ["llm"]
    },
    {
      "name": "qdrant",
      "status": "Healthy",
      "duration": 15.7,
      "tags": ["vector-db"]
    }
  ],
  "totalDuration": 81.7
}

📊 Logging

Logs are written to:

• Console: Colored output with timestamps
• Files: logs/enarro-[date].log (30-day retention)

Log levels can be configured per namespace in appsettings.json:

{
  "Serilog": {
    "MinimumLevel": {
      "Default": "Information",
      "Override": {
        "Microsoft.AspNetCore": "Warning",
        "Microsoft.EntityFrameworkCore": "Warning",
        "Enarro": "Debug"
      }
    }
  }
}

🧪 Development

Running Migrations

cd Enarro
dotnet ef migrations add MigrationName
dotnet ef database update

Building the Project

dotnet build

Running Tests

dotnet test

🐳 Docker Deployment

The application uses .NET Aspire for container orchestration. All containers are managed automatically:

• PostgreSQL: Port 5432 (persistent volume)
• Redis: Port 6379 (persistent volume)
• Qdrant: Port 6333 (HTTP), 6334 (gRPC) (persistent volume)
• Ollama: Port 11434 (persistent volume)

📈 Performance

• Concurrent Uploads: Up to 5 simultaneous document uploads
• Session Timeout: 60 minutes (sliding), 24 hours (absolute)
• Max File Size: 50 MB
• Chunk Size: 512 tokens with 50-token overlap
• Max History: 10 messages per session

🛣️ Roadmap

Phase 4: Enterprise Features (Planned)

• JWT Authentication & Authorization
• User Management System
• API Rate Limiting
• Advanced Query Rewriting
• Document Versioning
• Multi-tenancy Support

Phase 5: Optimization & Scaling (Planned)

• Hybrid Search (Semantic + Keyword)
• Re-ranking Pipeline
• Caching Strategies
• Performance Optimization
• Load Testing

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Microsoft Kernel Memory
• Ollama
• Qdrant
• .NET Aspire

📧 Contact

For questions or support, please open an issue on GitHub.

---

Built with ❤️ using .NET 10 and .NET Aspire