Lead Data Scientist - Safety Alignment

The Lead Data Scientist will design and implement safety architectures for Agentic AI systems, including guardrails, reward modeling, and self-monitoring capabilities. They will lead and collaborate on alignment techniques such as inverse reinforcement learning, preference learning, interpretability tools, and human-in-the-loop evaluation. Responsibilities include developing continuous monitoring strategies for agent behavior in both simulated and real-world environments, partnering with product, legal, Responsible AI, governance, and deployment teams to ensure responsible scaling and deployment, and contributing to and publishing novel research on alignment of LLM-based agents, multi-agent cooperation/conflict, or value learning. Proactively identifying and mitigating failure modes, such as goal misgeneralization, deceptive behavior, and unintended instrumental actions, and setting safety milestones for autonomous capabilities as part of deployment readiness reviews are also key duties.