====== Jerry Tworek ======
**Jerry Tworek** is an AI researcher and entrepreneur who previously served as Vice President of Research at OpenAI. In 2026, Tworek founded **Core Automation**, an artificial intelligence laboratory focused on developing advanced AI systems capable of automating aspects of AI research and development itself—conceptualized as "an AI to build AI."(([[https://www.therundown.ai/p/openai-reclaims-the-image-crown|The Rundown AI (2026]]))



===== Career at OpenAI =====
Tworek held a research leadership position at OpenAI, one of the leading organizations in large language model development and AI safety research. During his tenure at OpenAI, he contributed to the organization's research initiatives in machine learning and artificial intelligence advancement. His role as VP of Research placed him in a position to guide technical direction and oversee multiple research projects within the company.

===== Founding Core Automation =====
In early 2026, Tworek departed from OpenAI to establish Core Automation, a new AI laboratory with an explicit mission to develop AI systems capable of automating portions of the AI development pipeline. The founding team comprises experienced researchers and engineers drawn from major AI research organizations, including OpenAI, Anthropic, and DeepMind—three of the most prominent institutions in contemporary AI research and development.

The concept of "an AI to build AI" reflects a growing research direction within the field focused on automating aspects of model architecture search, hyperparameter optimization, training optimization, and other components of the AI development process. This approach aims to accelerate the pace of AI advancement by reducing manual engineering overhead in model development and refinement.

===== Team and Resources =====
Core Automation's founding team represents significant expertise aggregation from the AI industry's leading research organizations. The recruitment of researchers and engineers from OpenAI, Anthropic, and DeepMind indicates access to deep technical talent with experience in large-scale model training, alignment research, and AI safety—areas central to contemporary AI development challenges. This composition suggests the lab positions itself at the intersection of cutting-edge AI capabilities and systematic approaches to improving the AI development process itself.

===== Research Direction =====
The lab's focus on automating aspects of AI development aligns with broader industry trends toward improving efficiency in model training, architecture design, and optimization. Research areas of potential interest include neural architecture search (NAS), automated machine learning (AutoML), hyperparameter optimization, and meta-learning approaches that enable systems to improve their own learning processes. Such work could have substantial implications for accelerating AI research velocity and reducing computational requirements for model development.


===== See Also =====

  * [[core_automation|Core Automation]]
  * [[devin|Devin: Autonomous AI Coding Agent]]
  * [[recursive_superintelligence|Recursive Superintelligence]]
  * [[zac_hatfield_dodds|Zac Hatfield-Dodds]]

===== References =====