Amazon
Applied Scientist - Perception (SLAM/VIO), Fauna
$172K–$223K
New York, NY · Full-time · Posted Sep 1, 2026
- Experience
- 3+ years of building models for business application experience; PhD, or Master's degree and 4+ years of CS, CE, ML or related field experience.
- Education
- PhD, or Master's degree and 4+ years of CS, CE, ML or related field experience.
About the job
Develops and deploys real-time visual-inertial odometry, sensor fusion, and machine-learning-enhanced perception systems for robots, from research through embedded production deployment.
What you'll do
- Design and optimize VIO, state-estimation, sensor-fusion, and tracking algorithms for robotic platforms.
- Deploy and benchmark perception systems on resource-constrained hardware using real sensors and physical robots.
- Integrate learned perception techniques with geometric methods and collaborate across hardware, controls, and navigation teams.
What they ask for
- Hands-on experience developing and deploying Visual Inertial Odometry or visual-inertial SLAM systems; strong understanding of multi-sensor fusion and state estimation.
- Experience optimizing perception algorithms for embedded or resource-constrained hardware, with hands-on work using real sensor data, calibration, and physical robot platforms.
- Experience programming in Java, C++, Python or a related language, and in at least one listed area such as numerical optimization or high-performance computing.
- Experience in patents or publications at top-tier peer-reviewed conferences or journals.
Nice to have
- Experience with Unix/Linux, real-time systems, ARM/GPU platforms, ROS/ROS2, or embedded AI platforms such as NVIDIA Jetson or Qualcomm RB5.
- Experience shipping VIO or SLAM systems to production on physical robots, or integrating learned perception modules into geometric pipelines.
Pay
172,400.00 - 223,400.00 USD annually (base salary)
Tools
Java, C++, Python, Unix/Linux, ROS/ROS2, NVIDIA Jetson, Qualcomm RB5, ARM
AI focus
Applies learned features, depth estimation, neural odometry, and other ML-based perception methods alongside classical geometric techniques.
Model-written summary · the original posting is authoritative