Lumos Vector Core

Core abstractions and traits for the Lumos vector storage system. This crate provides the foundational types and interfaces that all vector storage implementations must follow.

Design Philosophy

This crate is inspired by the excellent designs of:

• Rig: Strong typing and trait-based abstractions
• Mastra: Declarative configuration and modularity
• Lumos: Enterprise features and performance

Core Concepts

• VectorStorage: The main trait for vector storage backends
• EmbeddingModel: Trait for embedding generation
• Document: Unified document representation with embedding support
• SearchRequest/Response: Structured query interface

Example

use lumosai_vector_core::prelude::*;
use std::collections::HashMap;

// Create a document with embedding support
let doc = Document::new("doc1", "This is a test document")
    .with_embedding(vec![0.1, 0.2, 0.3])
    .with_metadata("type", "article");

// Create an index configuration
let config = IndexConfig::new("test_index", 384)
    .with_metric(SimilarityMetric::Cosine);