Advance AI security and adversarial ML in mission-driven R&D: evaluate attacks and defenses for LLMs and agentic systems, develop interpretability and robustness methods, build evaluation frameworks and benchmarks, and translate insights for operational national-security deployment.