OpenAI is seeking a Research Engineer specializing in Multimodal Safety to join their team dedicated to advancing AI capabilities in seeing, hearing, speaking, and image creation. This role focuses on developing innovative safety techniques for state-of-the-art multimodal foundation models. The position involves conducting safety assessments, developing safety reward models, and implementing multimodal classifiers to ensure model compliance with safety protocols.
The ideal candidate will work at the intersection of multimodal research and safety, contributing to red teaming efforts and collaborating with cross-functional teams. Based in San Francisco with a hybrid work model (3 days in office), this role offers competitive compensation ($295K-$440K) plus equity and comprehensive benefits.
Key responsibilities include building evaluation pipelines, implementing risk mitigation techniques, developing moderation models, and working closely with safety teams to refine content policies. The role requires expertise in Python, multimodal research, and post-training techniques, along with strong collaborative skills.
OpenAI offers an inclusive work environment with excellent benefits, including comprehensive healthcare, mental health support, generous parental leave, and learning opportunities. This position presents a unique opportunity to shape the future of safe AI development while working with cutting-edge technology and contributing to OpenAI's mission of ensuring AI benefits humanity.