Skeg - The memory-efficient vector DB with high recall.

by
Skeg is a multi-tenant vector database built for extreme RAM efficiency: pack far more vectors per GB without trading away recall. Purpose-built for high-density, many-tenant deployments where memory is the bottleneck, not an afterthought. Open source, written in Rust.

Add a comment

Replies

Best
Hey everyone! 👋 Excited to finally launch Skeg on Product Hunt. We built Skeg because we were frustrated with the usual trade-offs in vector databases: either great recall but huge RAM usage, or low memory but poor performance under real load. Skeg is a disk-first vector database that focuses on memory efficiency while maintaining high recall. It works especially well in multi-tenant SaaS environments, RAG systems running alongside LLMs, and ARM-powered machines. Would love to hear your thoughts: What vector search tools are you currently using? What’s your biggest pain point with vector databases today? Happy to answer any questions! Repo: Thank you for checking it out!