Vector databases store high-dimensional embeddings and retrieve the nearest neighbours to a query vector, powering retrieval-augmented generation (RAG), semantic search and recommendations. Popular options include Pinecone, Weaviate and Qdrant.
Why teams use it
- Fast approximate nearest-neighbour search at scale
- The retrieval layer that grounds LLMs in your data
- Metadata filtering for hybrid semantic + keyword search
Install it
# Run Qdrant locally
docker run -p 6333:6333 qdrant/qdrant
pip install qdrant-clientA first working config
Create a collection and upsert embeddings.
from qdrant_client import QdrantClient
from qdrant_client.models import Distance, VectorParams, PointStruct
c = QdrantClient('localhost', port=6333)
c.recreate_collection('docs', vectors_config=VectorParams(size=1536, distance=Distance.COSINE))
c.upsert('docs', [PointStruct(id=1, vector=emb, payload={'source': 'runbook'})])Run and verify
hits = c.search('docs', query_vector=q_emb, limit=4,
query_filter={'must': [{'key': 'source', 'match': {'value': 'runbook'}}]})
for h in hits: print(h.score, h.payload)Mistakes to avoid early
- Mismatched embedding model between indexing and querying
- Ignoring metadata filtering, so results lack precision
- No re-embedding strategy when the model changes
Want senior help putting Vector Databases to work - or a team bootcamp on it? CodeWizard runs Israel's largest DevOps community and embeds named senior engineers with a transparent hours log. Book a free consult or explore the Academy.