Vector Databases: full course guide
Learn every major vector database — ChromaDB for local prototyping, Pinecone for managed production, Qdrant for advanced filtering, plus Weaviate, Milvus, and pgvector. One course, all the options.
Intermediate · read it in your browser with Skiiling Pro.
Loading the reader…
What is inside this guide
01Vector Search Fundamentals
- What are embeddings and why they matter
- Cosine similarity, dot product, Euclidean distance
- Dense vs sparse vs hybrid vectors
- ANN algorithms: HNSW, IVF, ScaNN
- Metadata filtering alongside vector search
- Chunking strategies that affect retrieval quality
02ChromaDB
- ChromaDB architecture: in-memory vs persistent
- Creating collections and adding documents
- Embedding functions: OpenAI, sentence-transformers
- Querying with metadata filters and where clauses
- Updating and deleting documents
- ChromaDB in a RAG pipeline (local dev)
03Pinecone
- Pinecone indexes: pod-based vs serverless
- Upserting vectors with metadata
- Namespaces for multi-tenant isolation
- Hybrid search: dense + sparse (BM25)
- Filtering with metadata at query time
- Pinecone with LangChain and LlamaIndex
- Production: batching, rate limits, monitoring
04Qdrant
- Qdrant collections and vector config
- Payload schema and indexed fields
- Advanced filtering: must, should, must-not
- Named vectors and multi-vector collections
- Scalar quantisation for memory efficiency
- Self-hosting Qdrant with Docker and Kubernetes
- Qdrant cloud vs self-managed trade-offs
05Weaviate, Milvus & pgvector
- Weaviate schema: classes, properties, modules
- Weaviate vectorizers: text2vec-openai, cohere
- GraphQL queries and hybrid search in Weaviate
- Milvus architecture: coordinator, data nodes, proxies
- Milvus collections, partitions, and indexes
- pgvector: enabling vector search in PostgreSQL
- pgvector index types: ivfflat vs hnsw
- Querying pgvector alongside relational data
06Choosing & Operating in Production
- Decision matrix: latency, scale, cost, filtering needs
- Embedding model selection and dimensionality trade-offs
- Index tuning: ef_construction, M, nprobe
- Re-ranking after vector retrieval
- Monitoring recall and search quality over time
- Data freshness: incremental upserts vs full re-index
- Security: API keys, VPC, encryption at rest
- Capstone: build a multi-source knowledge base with Qdrant
How to read it
The reader above opens the complete guide in your browser. The contents panel on the left is built from the guide's own bookmarks, so every module and topic is one click away. You can search the full text, zoom, and highlight passages in four colours; highlights and your reading position are remembered on this device.
Reading the guides is part of Skiiling Pro, the one web subscription that also unlocks Role Prep and Practical Labs and Skill Plans. The syllabus above is free to browse. See pricing.