Yongyuan Liang

Papers

Total Citations

H-Index

About

Yongyuan Liang is a rising researcher at the forefront of multimodal AI and agentic systems, with a focus on bridging digital and physical intelligence. Their most notable contribution is the development of **Magma**, a foundation model that extends traditional vision-language (VL) models to serve as a versatile AI agent capable of operating in both virtual and real-world environments. This work, published in 2025 and already garnering 19 citations, represents a significant leap forward by endowing VL models with the ability to not only understand multimodal inputs but also to take purposeful actions—a key step toward generalist embodied AI. Liang’s research addresses the critical challenge of integrating verbal intelligence with agentic capabilities, enabling systems to perceive, reason, and act across diverse domains. By pioneering models that unify understanding and action, Liang is helping to shape the next generation of AI assistants that can navigate complex tasks, from web browsing to robotic manipulation. Their work stands out for its practical ambition and technical rigor, marking them as a promising voice in the rapidly evolving field of multimodal agentic AI.