Target Audience

Data Scientists, Software Engineers, and Product Managers seeking to advance their AI capabilities with intelligent retrieval and agent-driven solutions.

Description

Explore how retrieval and agents come together to create “Smart RAG” systems that dynamically choose the right knowledge source for the task. Instead of treating retrieval as a black box, we’ll uncover the theory behind vectors, graphs, and agent orchestration.
You will:

• Understand the principles of Retrieval-Augmented Generation (RAG) and why it matters.
• Learn when vector search excels and when graph-based reasoning is necessary.
• Explore embeddings, chunking strategies, and other RAG pipeline stages
• Examine GraphRAG: extracting entities and relationships, designing graph schemas, and querying.
• Study agent architectures, tool design, and routing strategies for intelligent decision-making.
• See how these concepts connect into a complete “Smart RAG” system, from ingestion to deployment.

Main topics

• RAG by first principles:
  • limits of LLMs
  • embeddings & chunking
  • retrieval vs generation
  • graphs vs vectors
  • Smart RAG – Agent takes the wheel
• Vector DB and retrieval: motivation, basics, common tools
• Graph DB and retrieval: motivation, basics, common tools
• Agents for Smart RAG
  • Agentic Design Patterns
  • OpenAI Agents
  • tool design
  • decision traces
  • router strategies
• Workshop: building the Smart RAG app
  • System design walkthrough
  • pgvector + Neo4j
  • LlammaIndex
  • Agents with OpenAI agents
  • Gradio UI
  • Deployment (FastAPI endpoints, Dockerization, testing and demos)