Research Scientist, User Signals from Agentic Logs, DeepMind

DeepMind is seeking a Research Scientist to join the Gemini User Signals team in Zürich, Switzerland. This role focuses on enhancing the post-training strategy of advanced AI models by transforming user interactions and feedback into training signals that improve model behavior. Daily tasks include designing mechanisms to gather and utilize user signals, developing reward models, and creating post-training recipes applicable across various model types. The position requires expertise in reward modeling, preference optimization, and data-centric strategies, particularly in the context of Reinforcement Learning from Human Feedback (RLHF). Candidates should have a strong background in AI research and model evaluation. The advert does not specify language requirements or the working arrangement, but it emphasizes the collaborative nature of the work, making it suitable for individuals who thrive in interdisciplinary teams. This role is ideal for those passionate about advancing AI technology for public benefit and scientific discovery.