Show HN: NanoVector – A 120KB zero-dependency vector search engine in C and SIMD

A developer using the handle eminskinfo has published NanoVector, a 120KB vector search engine built in C and SIMD assembly, per a Show HN post on Hacker News. The project is hosted on GitHub and designed to perform vector similarity search without relying on any external dependencies.

Vector search engines are typically bulky libraries that power recommendation systems, semantic search, and similarity matching in applications. Most production engines (Weaviate, Milvus, Pinecone) run as separate services or require substantial runtime libraries. NanoVector aims to reduce that footprint dramatically by implementing core vector search operations in compiled C and using SIMD (single instruction, multiple data) instructions for performance on commodity CPUs.

The small footprint is the project's headline feature. At 120KB, NanoVector is orders of magnitude lighter than vectorized search libraries, making it a potential fit for embedded systems, edge devices, or applications where library size or external service dependencies are constraints. The zero-dependency design means developers can link it directly into their binaries without managing transitive dependencies or separate deployment. The use of SIMD allows it to exploit parallel CPU capabilities for faster distance calculations across high-dimensional embeddings, a core operation in vector search.

The project launched on Hacker News' Show HN section, where developers showcase personal projects and tools. At publication, the post had minimal engagement, with zero comments recorded, suggesting early visibility. The author provided the GitHub repository as the primary resource for documentation, source code, and any examples. No funding, team size, or commercial plans were announced.