Senior 3D Vision Scientist — Real-Time Robotics

Amazon; San Francisco (CA); On-site; USD 167,100 - 226,100; Full time

Job summary

Amazon Robotics is seeking a Senior Applied Scientist to advance 3D perception systems for real-time robotic platforms. You will develop scalable pipelines, optimize inference on ARM-based edge hardware, and deploy models that operate under strict latency constraints, contributing to perception, planning, and control in fulfillment centers.

The role emphasizes deep learning, transformer-based architectures, and large-scale data generation, with mentorship and collaboration across science,

Qualifications

• PhD or Master’s with 6+ years of applied research experience.
• 4+ years building machine learning models for business applications.
• Experience with 3D computer vision and robotics for real-time deployment.
• Proficiency in Java, C++, Python or related languages.

Responsibilities

• Architect, design, and implement 3D perception models for semantic occupancy prediction and scene completion on robotic platforms.
• Own end-to-end model lifecycle: scalable training, optimized inference latency on edge processors, production deployment.
• Design and scale pseudo-ground-truth data generation pipelines to produce training samples using SageMaker infrastructure.
• Drive multi-view perception integration by fusing multiple view camera inputs for robust 3D reconstruction.
• Mentor applied scientists and engineers, raise the technical bar, foster scientific rigor.

Skills

3D CV

Deep learning

Edge deployment

Python

C++

Education

PhD or MSc

Tools

ONNX

TensorRT

Job description

Amazon Robotics is seeking a Senior Applied Scientist to advance 3D perception systems for real-time robotic platforms. You will develop scalable pipelines, optimize inference on ARM-based edge hardware, and deploy models that operate under strict latency constraints, contributing to perception, planning, and control in fulfillment centers.

The role emphasizes deep learning, transformer-based architectures, and large-scale data generation, with mentorship and collaboration across science,