RE / RS - Foundations, Search
Core
Designing new embedding training objectives, scalable vector store architectures, and dynamic indexing methods to enable models to retrieve and condition on relevant information.
Role type
Researcher, embedding retrieval and vector search
Builds
Large-scale embedding systems and vector stores for OpenAI products and research
Domain
Artificial Intelligence, Machine Learning, Information Retrieval
Deliverable
production ML models
Required skills
representation learning, embedding models, vector retrieval systems, transformer-based LLMs, contrastive learning, metric learning, learning-to-retrieve systems, building and scaling large ML systems
Preferred skills
leading high-performance teams of researchers or engineers, first-principles mindset for retrieval and memory
Technologies
embedding models, vector store architectures, dynamic indexing methods, transformer-based LLMs
Responsibilities
Design new embedding training objectives and scalable vector store architectures; Drive innovation in dense, sparse, and hybrid representation techniques; Collaborate with Pretraining, Inference, and other Research teams to integrate retrieval throughout the model lifecycle; Tackle embedding models and retrieval systems optimized for grounding, relevance, and adaptive reasoning.
Seniority
Senior, hands-on IC
