Amazon is looking for talented Postdoctoral Scientists to join the Research and AI Development team at Amazon Robotics for a one-year, full-time research position with an optional extension for a second year. This Postdoctoral Scientist will innovate in the areas of multi-agent path planning, dynamic optimal transport (OT), and explainable AI (X-AI) for DeepFleet Foundation Models. They will have the opportunity to develop optimal and scalable solutions for the world's largest fleet of mobile robots, in addition to developing interpretability techniques for the DeepFleet FMs.
At Amazon, we experiment and innovate relentlessly. Science is core in our offering to shoppers, advertisers and customers. Our scientists apply machine learning, optimization, and probabilistic modeling at scale to enhance customer experience, help advertisers reach relevant audiences, and support brand building. We are seeking talented scientists to invent new techniques in a variety of areas and innovate on behalf of shoppers, advertisers, and customers.
Key job responsibilities
In this role you will:
A day in the life
On a typical day in this role, you will work to progress your research projects, meet with engineering, systems, and solutions stakeholders, brainstorm with other scientists on the team, and participate in team processes. You will follow your multi-agent path planning, dynamic OT, and X-AI projects through the entire life cycle of design, implementation, evaluation, analysis, and will communicate your findings and results through publications in top-tier academic venues.
About the team
The Research and AI Development team at Amazon Robotics is a multi-disciplinary science team that includes scientists with backgrounds in planning and scheduling, optimization, machine learning, and operations research. We develop novel planning algorithms and machine learning methods and apply them to real-word robotic warehouses, including: (1) Planning and coordinating the paths of thousands of robots (2) Dynamic allocation and scheduling of tasks to thousands of robots (3) Learning how to adapt system behavior to varying operating conditions and (4) Co-design of robotic logistics processes and the algorithms to optimize them.