Staff Research Scientist

Chewy Robotics is seeking a Staff Research Scientist to advance the orchestration of heterogeneous fleets of humanoid and mobile robots. The role involves leading research programs in multi-agent systems, optimization, and learning-based planning, with a focus on developing algorithms for fleet intelligence and real-world robotic applications.

Responsibilities:

• Lead research and technical strategy for one or more core fleet-intelligence areas, such as:Multi-agent perception and estimation
• Large-scale multi-robot task allocation and scheduling (LSTA)
• Multi-agent path finding (MAPF), routing, and congestion-aware coordination
• Fleet-level safety supervision and constraint-aware decision-making
• Learning-assisted planning and decision-making (RL/IL/model-based approaches)
• Research and design control methods for multi-agent robotic systems, such as collaborative and active SLAM, multi-robot task allocation, routing, estimation, and fleet-level decision-making under real-world constraints (battery, congestion, different sensing modalities, safety zones, SLAs)
• Design and run simulation-first experiments, define metrics for throughput, coordination latency/replanning responsiveness, robustness, and safety-event rate, comparing algorithms against strong baselines, instrumenting experiments, and analyzing fleet telemetry to understand performance and failure modes at scale
• Collaborate with scientists and engineers to develop pipelines for deploying the methods to real-world heterogeneous robotic systems
• Collaborate with scientists and engineers to integrate learning-based control into orchestration, working with RL/IL/model-based controllers for navigation, manipulation, or local behaviors and defining interfaces, rewards, and telemetry for end-to-end learning loops
• Build benchmarks and research artifacts, including reusable datasets, experiment harnesses, and publications for top-tier venues, while mentoring co-ops and junior contributors on related projects
• Contribute to the architecture and deployment of distributed robotic systems, including communication frameworks and APIs for full-scale deployment
• Utilize modern simulation and training environments, such as Isaac Lab or Gym for robot learning, and Isaac Sim, Gazebo, or Mujoco for robotic system modeling and validation