About Me

I’m a Senior Applied Scientist at Microsoft CoreAI, where I work on multi-agent systems, agentic AI, and foundation-model post-training.

Previously, I was a Senior Applied Scientist at Amazon, where I developed multimodal large language models and vision-language systems for video understanding and decision-centric applications, including action recognition, captioning, multimodal safety, and efficient video reasoning.

Earlier in my career, I was a Senior Research Scientist at Sony, developing real-time 3D character motion systems for interactive environments. At Tencent America, I worked on human pose estimation, video understanding, and motion intelligence. I began my industry research career as a postdoctoral researcher at Bosch, focusing on sensor-driven 3D hand gesture recognition.

My research focuses on decision-centric multimodal intelligence: building AI systems that can selectively perceive, efficiently reason, and reliably act. My current interests span LLM post-training, agentic AI, multimodal foundation models, and efficient multimodal reasoning, with the goal of bridging foundation models and real-world interaction to build scalable, grounded, and capable AI systems.

Photo: Mount Rainier — a reminder to navigate complexity, maintain perspective, and keep aiming higher.