VectorRAG.Net 0.1.18

VectorRAG.Net

High-performance embedded vector database for .NET 8.

VectorRAG.Net is an in-process vector database designed for Retrieval-Augmented Generation (RAG), semantic search, local AI assistants and low-latency applications.

Unlike external vector databases, VectorRAG.Net runs directly inside your application process.

No HTTP.

No Docker.

No external services.

No network latency.

Features

• SIMD-optimized vector search
• LSH-based ANN indexing
• Hybrid Search (Vector + BM25)
• Automatic document chunking
• Metadata filtering
• Snapshot save/load
• Runtime metrics
• Fully embedded architecture
• .NET 8 native implementation

Installation

dotnet add package VectorRAG.Net

Quick Start

var lshConfig = new EmbeddingLshConfig(
    Bands: 24,
    BitsPerBand: 12,
    MaxCandidates: 2048
);

var db = new VectorRAGDatabase(
    dimension: 1536,
    lshConfig: lshConfig
);

Add Documents

await db.UpsertTextDocumentAsync(
    externalId: "faq_001",
    text: File.ReadAllText("faq.txt"),
    metadata: new DocumentMetadata
    {
        Department = "Support"
    },
    embeddingModel: embeddingModel
);

Vector Search

var queryVector =
    await embeddingModel.GenerateEmbeddingAsync(
        "How do I reset my password?"
    );

var results = db.Search(
    queryVector,
    new SearchOptions
    {
        TopK = 5
    }
);

Hybrid Search

var results = db.Search(
    queryVector,
    new SearchOptions
    {
        TopK = 5,
        UseHybrid = true,
        TextQuery = queryText,
        Alpha = 0.7f
    }
);

Metadata Filtering

var results = db.Search(
    queryVector,
    new SearchOptions
    {
        TopK = 5,
        Filter = md =>
            md.Department == "Support"
    }
);

Save Snapshot

await db.SaveAsync("database.vdb");

Load Snapshot

await db.LoadAsync("database.vdb");

Typical Use Cases

RAG Systems

Store embeddings locally and retrieve context without external vector services.

Internal Knowledge Bases

Low-latency semantic search for documentation and support portals.

Desktop AI Applications

Run completely offline without Docker, cloud services or internet access.

Edge Computing

Vector search on embedded devices, industrial controllers and local servers.

Corporate Offline Systems

Suitable for environments where external network access is prohibited.

Performance

Example benchmark:

Actual performance depends on embedding dimensions, dataset size and hardware.

Licensing

VectorRAG.Net is commercial software.

Free Usage

The library may be used free of charge for:

• evaluation
• testing
• development
• research
• educational purposes
• proof-of-concept projects

No time limits.

No feature limitations.

Production Usage

Production use requires an active commercial subscription.

Current pricing:

• $1,000 USD per month per organization

For licensing questions:

Email: vipvodu@yandex.ru

Telegram: @vipvodu

Links

NuGet: https://www.nuget.org/packages/VectorRAG.Net

GitHub: https://github.com/likeslines-maker/VectorRAG.Net