Agility Robotics
Staff AI Engineer, Perception
$207K–$323K
Hybrid- Any Office (Fremont, CA, Salem, OR, or Pittsburgh, PA) +1 · Staff · Full-time · Posted Apr 23, 2026
- Location
- Hybrid- Any Office (Fremont, CA, Salem, OR, or Pittsburgh, PA)
Fremont, California, United States - Experience
- 5+ years deploying machine-learning-based object-detection algorithms on mobile robots.
- Education
- Master’s or Ph.D. in Artificial Intelligence, Robotics, Computer Science, or a related discipline.
- Visa sponsorship
- Applicants must have current authorization to work in the United States.
About the job
Owns the architecture and development of production object-perception systems for humanoid robots, including detection, tracking, scene understanding, and pose estimation.
What you'll do
- Set the architecture and technical roadmap for production object-perception systems.
- Design, develop, deploy, and optimize perception models for embedded systems.
- Collaborate with navigation, manipulation, and hardware teams to meet product requirements.
What they ask for
- 5+ years deploying machine-learning-based object-detection algorithms on mobile robots.
- Master’s or Ph.D. in Artificial Intelligence, Robotics, Computer Science, or a related discipline.
- Proficiency in deep convolutional neural networks, multi-object tracking, data association, supervised learning, and pose estimation; strong mathematics and computer-vision geometry fundamentals.
- Experience optimizing algorithms for CPU and GPU architectures and with MLOps; familiarity with computer-vision and machine-learning libraries.
Nice to have
- Publications in the field (CVPR, ICCV, RSS, ICRA preferred).
- Experience with YOLO, Faster/Mask R-CNN, 3- or 6-DoF pose-estimation models, or technical leadership.
Pay
$207,000 — $323,000 USD (base salary)
Tools
PyTorch, OpenCV, NumPy, YOLO, Faster/Mask R-CNN
AI focus
Develop and deploy robot perception models for multi-object detection and tracking, scene understanding, and 6-DoF pose estimation; evaluate state-of-the-art models and optimize neural networks for embedded systems.
Model-written summary · the original posting is authoritative