NanoVector
github.com/eminsk/nanovectorAbout
NanoVector is a roughly 120KB vector search and episodic-memory engine for AI agents, written in pure C99 with AVX2/FMA and ARM NEON SIMD optimizations and zero external dependencies. It targets the cold-start and dependency bloat of vector database clients like ChromaDB or FAISS, which can take 1.5 to 2.5 seconds to import and pull in hundreds of megabytes of dependencies, making it suited for CLI tools, serverless functions, and lightweight local agents.
Why it made the leaderboard
Gives agent builders a vector search and memory store that starts in milliseconds and adds no heavy Python ML dependencies, useful for CLI tools, serverless, and lightweight agent runtimes.
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.