Vector Embeddings and Semantic Search

Embeddings are dense vector representations that capture semantic meaning. Similar concepts map to nearby points in high-dimensional space.

Embedding Models

• OpenAI text-embedding-3-small: 1536d, cost-effective
• text-embedding-3-large: 3072d, higher accuracy
• Sentence Transformers (all-MiniLM-L6-v2): Open source, 384d, fast

Similarity Metrics

• Cosine Similarity: Measures angle between vectors (most common for text)
• Dot Product: Faster, works when vectors are normalized
• Euclidean Distance: L2 norm, less common for text

Vector Databases

Database	Type	Best For
Pinecone	Managed	Production, scale
Chroma	Open Source	Local dev, prototyping
Weaviate	Open Source	Complex queries
pgvector	PostgreSQL extension	Existing Postgres infra