Robotics Vision Scientist: VIO & Sensor Fusion

Amazon; New York (NY); On-site; USD 172,000 - 223,000; Full time

Job summary

Amazon is seeking an Applied Scientist to develop and optimize Visual Inertial Odometry (VIO) and sensor fusion systems for intelligent robots. You will own end-to-end pipeline from algorithm development to embedded deployment, ensuring perception runs efficiently on resource-constrained hardware.

You will apply modern ML approaches to enhance classical perception methods, prototype and test on real robots, and lead projects from research to production in a hands-on role.

Qualifications

• PhD or Master’s degree with 4+ years in CS/CE/ML.
• Experience or publications at top-tier conferences/journals.
• Proficiency in Java, C++, Python.
• Hands-on experience with Visual Inertial Odometry or SLAM.
• Strong multi-sensor fusion and state estimation knowledge.
• Experience deploying perception on embedded hardware.
• Experience with real sensor data and robotics platforms.

Responsibilities

• Design and implement VIO algorithms for real-time state estimation on robotic platforms.
• Develop multi-sensor fusion pipelines for accurate pose tracking.
• Optimize perception and tracking for embedded hardware with latency/power limits.
• Apply ML-based perception techniques to complement geometric estimation.
• Build and maintain calibration, evaluation, and benchmarking infra.
• Collaborate with hardware, controls, and navigation teams to integrate outputs.
• Lead technical projects from research prototyping to production deployment.

Skills

Java

C++

Python

Patents/Publications

Education

PhD or Master’s in CS/CE/ML

Tools

Unix/Linux

ROS/ROS2

Embedded hardware

Job description

Amazon is seeking an Applied Scientist to develop and optimize Visual Inertial Odometry (VIO) and sensor fusion systems for intelligent robots. You will own end-to-end pipeline from algorithm development to embedded deployment, ensuring perception runs efficiently on resource-constrained hardware.

You will apply modern ML approaches to enhance classical perception methods, prototype and test on real robots, and lead projects from research to production in a hands-on role.