Director, World Model & Agentic Learning

What you'll do

• Design systems for agents to represent and reason against accumulated domain knowledge rather than re-deriving information from raw sources.
• Build mechanisms for agentic learning that improve system performance through expert corrections, in-context learning, and outcome-driven refinement.
• Ensure all AI-generated conclusions are auditable, traceable, and reconstructable to maintain accountability in high-stakes environments.
• Partner with scientists and domain experts to ensure their expertise is integrated into the system's logic at scale.
• Recruit, develop, and lead a team of 4–8 AI scientists focused on knowledge representation and agentic systems.
• Establish technical direction and architecture for the enterprise world model and agentic learning capabilities.
• Collaborate with internal technology and evaluation teams to validate decision-quality improvements and integrate outcome signals.

What we're looking for

• Minimum 8 years of post-academic industry experience building and shipping AI/ML systems with ownership of technical architecture.
• Deep, hands-on expertise with large language models, retrieval-augmented generation, agentic frameworks, and knowledge representation.
• Demonstrated track record designing systems where knowledge accumulation, memory, or continual learning was the central technical challenge.
• Experience designing systems that improve from real-world operation and expert feedback through active learning or in-context learning.
• Strong people leadership experience including recruiting, building, and leading technical or scientific teams in a matrixed organization.
• Ability to set, defend, and hold a team accountable to a specific technical architecture.
• Advanced degree (PhD preferred) in computer science, AI/ML, applied mathematics, computational science, or a related discipline (preferred).
• Experience in regulated environments like life sciences or healthcare, and with knowledge graphs or ontologies (preferred).