Research Scientist, Socioaffective Agents

Research Scientists at Google DeepMind lead our efforts in developing novel algorithmic architecture towards the end goal of solving and building Artificial General Intelligence.

In this role, responsibilities will include making key contributions to Gemini workstreams, and pushing the frontier of agentic experiences.

Key responsibilities

• Prototyping: Developing new human-AI interaction experiences that waypoint and make tangible a positive future of relational and emotional AIs.
• Data: Unlocking new multimodal affective capabilities in large models, both pre-training and post-training, focusing on social agents.
• Models: Improving quality of models for understanding and generation. This may include research on improving tokenizers, training with implicit rewards, and/or general model behavior via fine-tuning.
• Evals: Better evaluation methods (human, auto raters, automated metrics) to measure quality of open-ended tasks. 

• PhD in Artificial Intelligence, Machine Learning, or related field.  Open also to backgrounds in Psychology (Social, Developmental, Cognitive), but must have strong Computational emphasis.
• Proven experience working with LLMs. Must have some multimodal experience, even if not a main focus.
• Proven track record of research and publications in one or more of the following areas: affective computing or sciences, LLM benchmarking, conversational design, audio/video generation.

• Experience of JAX, PyTorch or similar training frameworks
• Experience in Python and/or C++
• Experience applying and productionizing state-of-the-art multimodal research
• Diverse experience collaborating cross-function and with other researchers

The US base salary range for this full-time position is between $166,000 - $244,000 + bonus + equity + benefits. Your recruiter can share more about the specific salary range for your targeted location during the hiring process.