USearch: Smaller & Faster Similarity Search & Clustering Engine

Portable single-header similarity search and clustering library with SIMD-optimized and user-defined metrics, supporting 15+ quantization levels from f64 to f6 and single-bit, with native bindings for C++, Python, Rust, JavaScript, Java, Go, Swift, C#, C, Objective-C, and Wolfram, trusted by Google, Microsoft, ClickHouse, DuckDB, ScyllaDB, Yugabyte, TiDB, and others.