Familiarity with AI and LLM observability platforms such as LangSmith, Langfuse, Arize Phoenix/AX, Datadog LLM Observability, MLflow, Galileo, Fiddler, AgentOps, or equivalent tools, including understanding traces, agent steps, tool calls, latency, errors, token use, cost, response quality, retrieval quality, guardrails, and outcome monitoring. Familiarity with agentic AI frameworks and components such as LangGraph, Semantic Kernel, AutoGen, CrewAI, or equivalent orchestration approaches, with the ability to credibly discuss architecture, multi-agent workflows, tool calls, memory, context, and human approval points.