Description
The purpose of this training is to teach how to generate embeddings from text and images, store them in vector databases (FAISS, Chroma, PgVector, Milvus, Weaviate), perform efficient similarity search, and integrate them with RAG scenarios in production environments. Participants will gain hands-on experience with chunking strategies, metadata schemas, index and distance metric selection, hybrid search, and performance evaluation.
Learning Outcomes
- Generate embeddings and store them in Oracle 23ai, Chroma, PgVector, and Milvus
- Design hybrid search and re-ranking pipelines
- Build and optimize RAG workflows for quality and performance
Audience
- Backend and data engineers
- Search and RAG application teams
- Teams building semantic search and retrieval systems
Outline
Vector & Retrieval Fundamentals
- Embedding models and normalization
- Chunking strategies
- Distance metrics and indexing
- Vector database types
- Metadata schema design
- Hybrid search and evaluation
- Hands-on: PgVector similarity search
RAG Integration & Best Practices
- RAG architecture
- Advanced chunking strategies
- Re-ranking techniques
- Performance and cost optimization
- Observability and governance
- Hands-on: End-to-end RAG pipeline
Prerequisites
Basic programming knowledge and familiarity with data structures.