Post-Doctoral Researcher, PhD, DeepMind, 2026 Start (Fixed-Term Contract)

Minimum qualifications:

• PhD degree in a STEM field (e.g., Computer Science, Mathematics, or Statistics) or equivalent practical experience.
• Experience with one or more general purpose programming languages: Python, C/C++, or Java, etc. and training deep learning models in frameworks such as JAX, TensorFlow, or PyTorch.
• Research experience in machine learning or AI techniques (e.g., open source project(s), campus lab experience, research internship(s), or publication(s)).
• One or more scientific publication submission(s) in AI conferences (e.g., NeurIPS, CVPR, ICCV, ICLR).

Preferred qualifications:

• PhD and Master’s degree in a STEM field (e.g. Computer Science, Mathematics, or Statistics) or equivalent practical experience.
• Hands-on experience with one or more general purpose programming languages: Python, C/C++, or Java, etc and training deep learning models in frameworks such as JAX, TensorFlow, or PyTorch.
• Extensive research experience in machine learning or AI techniques (e.g. open source project(s), campus lab experience, research internship(s), or publication(s)).
• Ability to work and collaborate across multiple teams and work with ambiguity.
• A strong publication record in top-tier conferences/journals.

About the job

The Post-doctoral Researcher role is a 12 to 24-month role designed to explore challenging problems for pushing frontiers of AI, while collaborating with DeepMind scientists and engineers. Your role will entail conducting human-computer interaction (HCI) research on how people interact with, oversee, and provide feedback to increasingly autonomous AI agents, designing and building novel interfaces and interaction techniques for human-agent collaboration, and evaluating them through rigorous user studies. Insights from your work will inform how agents learn from human feedback and remain safe and aligned with human intent, with potential application to products that might have a positive impact on billions of users.

At DeepMind, we're a team of scientists, engineers, machine learning experts and more, working together to advance the state of the art in artificial intelligence, and fulfill our mission of building AI responsibility to benefit humanity. We use our technologies for widespread public benefit and scientific discovery, and collaborate with others on critical challenges, ensuring safety and ethics are the highest priority.

In particular, our AQUA team at DeepMind India works at the frontier of AI agents, focusing on autonomous self-improvement, reinforcement learning, and the safety and alignment of agents. We build agents that continually improve themselves by learning from their own experience, interactions, and feedback, and we develop methods to ensure that increasingly capable and autonomous agents remain safe, reliable, and aligned with human intent and values. We aim to bring our research into Google's cutting-edge models and products, while also disseminating our findings through publications at top ML conferences/journals.

Responsibilities

• Work with research mentors to formulate research projects or novel applications of machine learning aligned with the team's mission. Conduct research and publish high quality work.
• Design and execute large-scale experiments, writing high-quality, reusable code, and contributing to a production-ready system using frameworks such as JAX. Learn and understand research in machine learning algorithms.
• Understand how to use research to drive design decisions. Design and conduct user studies (e.g. interviews, surveys, controlled experiments, and field deployments) to understand how people interact with, oversee, and provide feedback to AI agents.
• Prototype novel interfaces and interaction techniques that help people steer, supervise, and collaborate with increasingly autonomous agents. Develop human-centered evaluation methods and metrics for the helpfulness, trustworthiness, safety, and alignment of agents.
• Translate insights from human studies into improvements in how agents learn from human feedback and improve themselves (e.g. reward modeling, preference elicitation, and human-in-the-loop reinforcement learning).