Sangha Partners

Compensation

Full Description

Our client is building advanced robotic systems that operate in real industrial environments, where reliability, perception accuracy, and precise control matter every second. As a Machine Learning Engineer focused on robotics perception and manipulation, you’ll develop the ML-driven vision and control systems that enable robots to understand their surroundings, identify weld seams or work surfaces, and execute complex tasks with precision. This is not a generic ML role. This is not NLP or LLM work. This is real robotics- object detection, depth, segmentation, and closed-loop control running on physical hardware in production environments. If you’ve trained and deployed perception models onto robots, we want to talk to you. What You’ll Do • Build real-time perception pipelines for object detection, segmentation, depth estimation, and geometric understanding in challenging industrial settings • Develop manipulation and control policies that integrate perception signals for trajectory generation, tool control, and servoing • Own end-to-end ML workflows, from data collection to model training, optimization, deployment, validation, and continuous improvement • Integrate ML models with physical robot hardware (robot arms, sensors, depth cameras, calibration pipelines) • Debug real systems: camera extrinsics, sensor alignment, latency bottlenecks, failure modes • Collaborate closely with robotics, controls, and field engineering teams to achieve scalable, stable performance in the real world • Contribute to on-site testing, iteration, and validation in customer environments What You Bring • 3–7+ years of experience building ML systems for robotics, autonomous systems, or real-world computer vision • Hands-on experience with object detection, segmentation, depth estimation, or tracking (e.g., YOLO, Detectron2, segmentation networks, 3D CV) • Experience deploying ML/CV models on real robots or hardware, not just simulation • Strong coding skills in Python and C++ for real-time robotics applications • Experience working with RGB-D, stereo, or structured-light cameras, including calibration and debugging • Strong grasp of robotics fundamentals: perception → planning → control loop • Experience optimizing models for real-time inference (<50–100 ms constraints) Strongly Preferred • Experience with robot arms (ABB, Fanuc, UR, KUKA, Yaskawa, etc.) • Familiarity with ROS/ROS2, hardware bring-up, and sensor integration • Background in industrial robotics, automation, or manufacturing environments • Experience with SLAM, 3D reconstruction, point clouds, or visual servoing • Experience deploying systems in harsh, cluttered, or dynamic environments Why This Role Is Unique • Your models are deployed on real robots, not in notebooks • Rapid iteration → what you build ships to production quickly • Huge ownership over perception, control, and deployment pipelines • Work directly with field teams to solve real industrial challenges • Opportunity to shape the core ML stack of a frontier robotics platform