About Us
Our team drives the Personal Intelligence research behind Gemini, with a mission to make AI more personal, proactive, and context-aware. You will push the boundaries of Large Language Models (LLMs) to build the brain of the world’s most helpful personal assistant—one that securely integrates with users' personal data to solve real-world problems. Our core research focus areas include:
- Personalization: Inferring and adapting to user intent, preferences, communication styles, etc.
- Context: Reasoning over extensive personal history (e.g., Gmail, Photos, Drive) across diverse modalities (Text, Image, Audio, Video).
- Agency: Empowering AI models to autonomously plan, use tools, and execute complex tasks on behalf of the user.
The Role
As a Research Scientist for Gemini Personal Intelligence, you will advance the state-of-the-art in understanding and reasoning to create an AI that truly understands, remembers, and adapts to the user's unique life and context. Key responsibilities for this role:
- Driving research on post-training techniques (e.g., RL, SFT, and preference optimization) specifically tailored for personalization scenarios.
- Developing novel evaluation frameworks and simulation methods to measure model quality against user behaviors / feedback.
- Designing and training agents capable of orchestrating tools and APIs to deliver hyper-personalized experiences.
About You
We are seeking a Research Scientist who can drive new research ideas from conception and experimentation through to productionisation. In this rapidly shifting landscape, we regularly invent novel solutions to open-ended problems. You should be flexible, adaptable, and comfortable pivoting when ideas don’t work out.
In order to set you up for success as a Research Scientist at Google DeepMind, we look for the following skills and experience:
- PhD in Machine Learning, Computer Science, or a relevant field (or equivalent practical research experience).
- A proven track record of research excellence (e.g., publications at top-tier venues like NeurIPS, ICML, ICLR, or significant industry contributions), ranging from recent graduates to experienced researchers.
- Strong software engineering skills to complement your research background.
In addition, the following would be an advantage:
- Hands-on experience with modern post-training methods (SFT, RLHF, etc.).
- Prior work applying LLMs to personalization, memory, or agentic workflows.