AI Security & Red Teaming: Master direct/indirect attacks, RAG poisoning, AI agent risks, and defense-in-depth.