Qdrant is a Rust-based vector database built for production RAG systems that need more than basic similarity search. It shines when you need hybrid search with metadata filtering (like finding similar docs but only from a specific category), or when you're scaling beyond a single machine with sharding and replication. The filtering is genuinely flexible, you can query by any payload field during vector search, not as a post-processing step. It supports multiple vector types per point, which matters if you're doing dense and sparse retrieval together. If you're prototyping or don't need filters, Chroma is simpler. If you want zero ops, use Pinecone. But if you're building something real and want control over your infrastructure, Qdrant is solid.
npx skills add https://github.com/orchestra-research/ai-research-skills --skill qdrant-vector-search