Light Origins builds foundation models for Physical AI, extending foundation model intelligence from the digital world into the physical world.
AI is moving from understanding the world to interacting with it. The next challenge is not only more capable models, but AI that can operate safely and reliably in the real world - with outcomes that can be verified.
Light Origins takes an intelligence-first approach. Our work spans foundation models, multimodal perception, real-world feedback, robotic systems, and continual learning.
One system. Models, compute, data, hardware, and deployment — all feeding the same learning loop.
Roger Jiang is the founder and CEO of Light Origins. Previously at OpenAI, he contributed across pre-training, alignment, optimization, and infrastructure for InstructGPT, ChatGPT, and GPT-4.
His work included foundational RLHF and InstructGPT research, large-scale model training, and FP8 training and inference. He is credited eight times in the GPT-4 Technical Report.
Roger holds a Ph.D. in Physics from the University of Maryland.