Bingyi Kang, an AI researcher focused on building intelligent agents for the physical world, has emerged as a key figure in the fast-growing field of spatial intelligence. Now a Founding Scientist at AMI Labs, Kang is leading efforts to develop systems that can both perceive and act in real-world environments—an ambition that sits at the core of next-generation artificial intelligence. His work spans computer vision, reinforcement learning, and multimodal AI, with a particular emphasis on enabling machines to understand space with human-like precision.
Kang is best known as the creator and leader of the Depth Anything series, a family of models designed to advance depth perception in AI systems. The project, which includes iterations such as DAv2, PromptDA, and Video Depth Anything, has gained widespread adoption among researchers and developers, collectively attracting more than 24,000 GitHub stars. By focusing on scalable, general-purpose depth estimation, Kang’s work addresses a critical bottleneck in robotics, autonomous systems, and augmented reality—domains where accurate spatial understanding is essential. His contributions have helped position depth perception as a foundational capability for embodied AI.
Before joining AMI Labs, Kang led the Spatial Intelligence research team at ByteDance Seed, where he worked on pushing the boundaries of multimodal reasoning and perception at scale. His broader career includes experience across leading AI institutions, including Sea AI Lab, Facebook AI Research (FAIR), UC Berkeley, and the National University of Singapore, where he completed his PhD. Known for bridging academic research with practical deployment, Kang represents a new generation of AI scientists focused on building systems that move beyond static understanding toward real-world interaction. His work signals a broader industry shift toward intelligent agents capable of seeing, reasoning, and acting in complex environments.