Research
Research identity and directions for AI evaluation, agent behavior, data production, and AI-native organizations.
Agent Evaluation
Designing evaluation systems for agents that plan, call tools, collaborate with humans, and operate across messy real-world tasks. The focus is on traces, judgment quality, failure modes, and durable feedback loops.
Human-Agent Collaboration
Studying how humans and agents divide attention, authority, memory, and decision rights inside practical work. The question is not replacement, but how new forms of cooperation become legible and trustworthy.
Data Strategy
Treating data production as a strategic system: task design, labeling operations, expert review, synthetic data, evaluation data, and the organizational habits that determine data quality over time.
AI-Native Organizations
Exploring how organizations change when AI systems become part of coordination, learning, product development, and management. The unit of analysis is the workflow, not the slide deck.