Ted Staley is a senior engineer at the Johns Hopkins Applied Physics Lab, where he works in in the research and development department. Ted started his career in robotics, and began working with deep learning systems as a solution to robotic control (deep reinforcement learning). With the rise of large transformer models, he shifted his attention to LLMs and subsequent developments in multimodal models (VLMs, VLAs) as promising paradigms for more general AI systems. His primary interests lie in how these techniques can be further generalized and potentially adopted for general purpose robot systems.
In 2024 and 2025 Ted helped lead efforts at APL to build custom LLMs from the ground up, with an eye towards bespoke AI models for future government needs. In these ongoing projects, Ted is responsible for dataset processing and the distributed training routines of LLMs at the scale of a few billion parameters, with datasets in the few trillions of tokens. Currently he is focusing on how to expand this capability at APL to the broader systems described above.
Education History
- B.S., Mechanical Engineering, Johns Hopkins University
- M.S., Robotics, Johns Hopkins University
Work Experience
Senior Professional Staff, JHU Applied Physics Laboratory