ABOUT THE TEAM
The Foundations Research team works on high-risk, high-reward ideas that could shape the next decade of AI. Our goal is to advance the science and data that enable our training and scaling efforts, with a particular focus on future frontier models. We push the boundaries of data, scaling laws, optimization techniques, model architectures, model + search co-design and efficiency improvements to propel our science.
The Search team sits within Foundations, building agentic search by co-designing model-system interfaces with the core search stack: serving, indexing, and retrieval. We research and develop large-scale systems that transform and index vast corpora, enabling models to most effectively find and reason over information.
ABOUT THE ROLE
We’re looking for a Researcher to work on semi-structured knowledge graphs within Search. You’ll research and develop methods for mining structured information from heterogeneous, unstructured and semi-structured web content, making that information more accessible and useful to AI models.
The core questions are: What structure helps models find and use information? How can we extract it reliably at web scale? And how do we measure whether it improves the model’s ability to reason and answer questions? How do we design the system to be optimizable end-to-end?
You’ll work across research, data, and systems, exploring how advances in AI and compute can improve the way knowledge is extracted, organized, and retrieved. This work will support retrieval across OpenAI products and internal research efforts, with opportunities for scientific publication.
RESPONSIBILITIES
- Develop methods to extract and connect structured information across diverse web sources, including entities, attributes, and relationships.
- Build scalable approaches for transforming noisy, inconsistent content into useful knowledge representations.
- Explore how models and compute can improve extraction quality and reduce the manual work involved in building knowledge systems.
- Design evaluations that measure both the quality of extracted knowledge and its usefulness for retrieval and model reasoning.
- Partner with researchers and engineers to integrate structured knowledge into Search and use model performance to guide improvements.
YOU MIGHT THRIVE IN THIS ROLE IF YOU HAVE
- Deep experience in knowledge extraction, knowledge graphs, entity resolution, or related work that derives structure from large, heterogeneous datasets.
- Strong machine learning and research skills, with experience developing and evaluating methods for real-world data.
- A track record of building or scaling data and ML systems, particularly for web-scale extraction, indexing, or retrieval.
- Experience independently driving open-ended technical problems from an initial idea through experimentation and implementation.
- A first-principles approach to using AI and compute, with a willingness to rethink established methods as model capabilities improve.
- A strong interest in how the representation of information shapes what models can find, understand, and use.
About OpenAI
OpenAI is an AI research and deployment company dedicated to ensuring that general-purpose artificial intelligence benefits all of humanity. We push the boundaries of the capabilities of AI systems and seek to safely deploy them to the world through our products. AI is an extremely powerful tool that must be created with safety and human needs at its core, and to achieve our mission, we must encompass and value the many different perspectives, voices, and experiences that form the full spectrum of humanity.
To notify OpenAI that you believe this job posting is non-compliant, please submit a report through this form https://form.asana.com/?d=57018692298241&k=5MqR40fZd7jlxVUh5J-UeA. No response will be provided to inquiries unrelated to job posting compliance.
At OpenAI, we believe artificial intelligence has the potential to help people solve immense global challenges, and we want the upside of AI to be widely shared. Join us in shaping the future of technology.
