Research Portfolio

Two Threads, One Question.

How do we build AI systems that truly understand human intent? I approach this question from two complementary angles: formalizing the evaluations AI must master, and designing the interaction paradigms that keep humans in control.