Towards Data Science is a community-powered publication that showcases work in data science, machine learning and artificial intelligence. Every day newcomers, seasoned researchers and industry practitioners publish tutorials, research notes and real-world case studies that help the field move forward.

Towards Data Science

For small to medium-scale RAG systems, NumPy and SciKit-Learn can handle vector search without dedicated vector databases. The article demonstrates building a production-ready retrieval component using matrix multiplication for cosine similarity, showing how to search millions of text chunks in milliseconds using only in-memory operations. It includes working code for both NumPy-based brute-force search and SciKit-Learn's optimized nearest neighbor algorithms, with performance benchmarks on 1.2 million lines of text. Vector databases become necessary only when you need persistence across restarts, RAM exceeds available memory, frequent CRUD operations, or complex metadata filtering.

You Probably Don’t Need a Vector Database for Your RAG — Yet