We study the conditions that make AI systems reliable, useful, and honest. Our research is driven by application: it begins with real deployment challenges and produces findings that govern what we build.
Applied research begins with choosing the right problems. Each area below addresses the gap between what AI systems can do and what they reliably do: a gap that carries real-world consequences.
AI systems do not fail only at the edge. We examine how they behave under real-world conditions: ambiguity, failure, drift, and misuse. Reliability is not what a model does in isolation. It is what it continues to do in practice.
Our work examines how systems access, structure, and retrieve the knowledge they depend on. Grounding is not only about finding information. It is about making what a system knows accurate, current, and usable.
We study how AI systems retain context, form useful memory, and adapt across time and interaction. Intelligence does not begin again with every prompt. It accumulates through continuity.
What we study shapes what we build and deploy: how systems behave, where they fail, and what they require to work in the real world. Our products, infrastructure, and solutions are not separate from our research. They are its continuation.