Build Responsible AI controls for content safety, bias, privacy, transparency, explainability, prompt injection, jailbreaks, hallucination, data leakage and inappropriate model behavior, using guardrail frameworks such as NeMo Guardrails, Guardrails AI or the native guardrails from each cloud provider. The architect is expected to write production-grade Python, build agentic workflows in LangGraph and LangChain, instrument solutions with Langfuse, and personally build proofs of concept, reference implementations and core framework components.