🔬 Research
I focus on building systems that can interpret messy, real-world input and respond in ways that are useful beyond narrow training settings. My work explores how perception, structure, and learned behavior can come together to produce models that adapt to new goals, environments, and demands—without needing to be reprogrammed.
I use simulation and scene understanding not as ends in themselves, but as scaffolds for learning behavior that holds up under ambiguity and change. The goal is simple: to design systems that remain coherent and capable when the world doesn't follow a script.
World Understanding
Constructing structured representations of the world - visual, spatial, and semantic - from raw sensory input. This includes understanding object permanence, affordances, and 3D spatial structure in ways that support planning and adaptation.
Key Techniques
Goal Grounding
Translating intent - expressed through language, demonstrations, or observation - into internal objectives that machines can reason about and act upon. This forms the bridge between human intention and machine execution.
Key Techniques
Structured Generalization
Designing systems that extend what they've learned to new tasks, domains, or goals by leveraging internal structure - semantic, spatial, or temporal. Including architectures that support adaptation, reuse, and abstraction without brittle retraining.
Key Techniques
Simulated Generalization
Using simulation not just as a training ground, but as a tool for scalable data generation, safe testing, and robust transfer to the real world. This includes planning through simulation and real-time feedback loops.
Key Techniques
Let's Connect
Interested in collaborating on AI research, discussing opportunities, or just connecting? I'd love to hear from you.
Email Me
Drop me a line for research collaborations, opportunities, or just to say hello.
Schedule a Call
Let's have a conversation about AI, research, or potential collaborations.
Book a Meeting