Design, build, evaluate, and productionize agents in Azure AI Foundry and Copilot Studio: tool and function calling, orchestration and multi-agent patterns, grounding and retrieval over Dataverse and document stores, prompt and context engineering, structured output, and human-in-the-loop checkpoints. • Implement responsible AI controls appropriate to a child welfare setting: content safety, PII/PHI handling, bias and fairness testing, explainability, audit logging of AI-influenced actions, and clear framing of AI output as decision support rather than decision making.