Summary
Weicheng Kuo is a Staff Research Scientist at Google DeepMind with a decade of experience in computer vision and machine learning, transitioning from a Research Scientist role at Google AI where he contributed to CVPR/ICCV/ECCV/NeurIPS publications and product integrations like Maps, Waymo, and Vertex AI. His work spans multimodal post-training (SFT, RL), spatial understanding (detection, segmentation), STEM education applications, and multimodal safety, reflecting a rare combination of foundational research and product-minded engineering. He earned advanced training at UC Berkeley and has interned on video analysis and dereflection projects at Apple and Google, respectively. Colleagues describe him as someone who moves seamlessly between cutting-edge models and practical deployment, and he maintains a personal research webpage showcasing his work. Notably, his current focus on multimodal post-training for Gemini demonstrates an emphasis on aligning large models for safe, spatially-aware applications.
10 years of coding experience
5 years of employment as a software developer
PhD student Computer vision and machine learning, PhD student Computer vision and machine learning at University of California, Berkeley
NTU