Yao (Jason) Lu is a principal research scientist at NVIDIA Research. His current research interest is efficient large language models (LLM) and vision language models (VLM). Before joining NVIDIA, he was a TLM at Google Deepmind where he worked on reinforcement learning, imitation learning on embodied AI. He co-led the SayCan, RT-1, RT-2, and RT-X algorithms that have been featured extensively by media, such as New York Times, Washington Post, Forbes, Reuters, TechCrunch, The WIRED, and so on.