About the job
Join the Personal Robotics Group at Amazon, where you'll help pioneer intelligent robotic products that deliver meaningful customer experiences. As a Senior Applied Scientist, you'll drive technical excellence in robotic task planning, developing and improving algorithms that enable robots to learn and execute complex tasks in dynamic real-world environments, while establishing evaluation and benchmarking methodology to measure system performance.
Responsibilities
Research and develop novel approaches to robotic task planning, task understanding, and reasoning to improve system performance
Design and implement evaluation frameworks and benchmarks for robotic planning systems, driving measurable improvements
Explore and fine-tune models to enhance robot capabilities across diverse tasks and environments
Build and maintain datasets and data collection pipelines for training and evaluating models
Define quantitative metrics and success criteria that connect evaluation results to real-world robot performance
Collaborate with engineering teams to integrate research findings and evaluation into development workflows
Partner with cross-functional teams to ensure holistic improvement of robot capabilities
Stay current with the latest advancements in robotics, task planning, and AI research
Qualifications
Minimum
PhD, or Master's degree and 6+ years of applied research experience
3+ years of industry or academic research experience
Experience with programming languages such as Python, Java, C++
Experience designing experiments with quantitative evaluation and benchmarking methodology for AI or robotics systems
Experience with LLMs, VLMs, or large multimodal models and their evaluation challenges
Experience with robotics task planning, reasoning, or complex task understanding
Preferred
Strong publication record at major conferences (RSS, CoRL, ICRA, IROS, HRI, NeurIPS, ICML, or similar)
Experience building evaluation pipelines or benchmarks for robotics or AI systems
Experience with model fine-tuning and training for robotics applications
Experience with complex task decomposition and multi-modal task understanding
Experience with simulation environments for robotics
The ability to work with minimal guidance, be proactive, deal with ambiguity, and thrive with quickly evolving goals
Demonstrated experimental mindset with a track record of hypothesis-driven research
Experience bridging research with practical engineering implementation in robotics systems