FluxIndex.SDK 0.2.10

FluxIndex

.NET 9.0 RAG library for building retrieval-augmented generation systems with vector and hybrid search.

Overview

FluxIndex provides document indexing and semantic retrieval for .NET applications. It combines vector embeddings with keyword search using clean architecture principles and modular design.

Features

• Vector Search - Semantic similarity using embeddings (SQLite-vec, pgvector)
• Keyword Search - BM25 algorithm for exact term matching
• Hybrid Search - Combines vector + keyword with Reciprocal Rank Fusion
• Adaptive Search - Automatic strategy selection based on query complexity
• Embedding Cache - In-memory cache for repeated queries (100% improvement)
• Storage Options - SQLite (dev), PostgreSQL (production)
• AI Agnostic - Works with OpenAI, Azure OpenAI, or custom services
• Document Processing - FileFlux integration for PDF/DOCX/TXT
• Web Crawling - WebFlux integration for web content extraction
• Semantic Caching - Redis-based similarity caching
• Clean Architecture - Modular DI-based design

Installation

# Minimal setup (local development)
dotnet add package FluxIndex.SDK
dotnet add package FluxIndex.Storage.SQLite

# Production setup
dotnet add package FluxIndex.SDK
dotnet add package FluxIndex.AI.OpenAI
dotnet add package FluxIndex.Storage.PostgreSQL
dotnet add package FluxIndex.Cache.Redis

# Document processing (optional)
dotnet add package FluxIndex.Extensions.FileFlux
dotnet add package FluxIndex.Extensions.WebFlux

Quick Start

using FluxIndex.SDK;

// Setup with SQLite and OpenAI
var context = FluxIndexContext.CreateBuilder()
    .UseSQLite("fluxindex.db")
    .UseOpenAI("your-api-key", "text-embedding-3-small")
    .Build();

// Index documents
await context.Indexer.IndexDocumentAsync(
    content: "FluxIndex is a .NET RAG library for semantic search.",
    documentId: "doc-001"
);

// Search with adaptive strategy
var results = await context.Retriever.SearchAsync(
    query: "RAG library for .NET",
    maxResults: 5
);

foreach (var result in results)
{
    Console.WriteLine($"Score: {result.Score:F2}");
    Console.WriteLine($"Content: {result.DocumentChunk.Content}");
}

Performance

Based on benchmarks with .NET 9.0 on Intel i7-1360P:

Optimization Features:

• In-memory embedding cache for repeated queries
• Semantic caching with Redis for similar queries
• Optimal parallelism: 8 threads for batch operations
• Expected 30% latency reduction with realistic cache hit rates

Advanced Usage

See Tutorial for detailed examples:

• Hybrid Search - Combine vector and keyword search with RRF
• Adaptive Search - Automatic strategy selection based on query complexity
• Embedding Cache - In-memory cache for repeated queries
• Semantic Caching - Redis-based similarity caching for related queries
• Document Processing - FileFlux integration for PDF/DOCX/TXT
• Web Crawling - WebFlux integration for web content extraction
• Batch Operations - Efficient bulk indexing with parallelism

Documentation

• Getting Started - Setup and configuration
• Tutorial - Comprehensive examples
• Architecture - Design principles
• Cheat Sheet - Quick reference

Examples

• RealWorldDemo - OpenAI + SQLite integration
• FileFluxSample - PDF/DOCX processing
• WebFluxSample - Web crawling

Requirements

• .NET 9.0 or later
• SQLite or PostgreSQL
• OpenAI API key (optional)

License

MIT License - see LICENSE file.

Contributing

See development roadmap for planned features.