Senior Machine Learning Engineer - Computer Vision for Robotics

The AI Teams at Agile Robots are looking for a Senior Machine Learning Engineer (m/f/d) Computer Vision for Robotics, who will design and train deep-learning models for visual servoing in precision robot manipulation. Your model's output steers the robot as it moves, so inference latency and sub-millimetre accuracy both decide whether the task succeeds.

Responsibilities

• Method Development: Develop deep-learning methods for visual servoing and robotic vision, adapting and combining existing methods for new tasks where that requires significant changes.
• Model Training: Design, train, fine-tune, and evaluate models in PyTorch, including adapting pretrained and foundation models to real robot tasks.
• Perception Methods: Develop 2D and 3D perception for pose estimation, depth, tracking, and geometric understanding, combining learned, classical, and optimization-based methods where each fits.
• Closed-Loop Integration: Integrate learned models into closed-loop robot control for high-precision manipulation.
• Robot Evaluation: Evaluate models on real robots and precision manipulation tasks, analyse failure cases, and improve robustness and generalization.
• Real-Time Deployment: Build real-time inference pipelines in C++ and Python and optimize models with ONNX, TensorRT, and CUDA to meet the timing of the control loop.

Essential Skills

• Background: Master's degree or PhD in computer science, robotics, electrical engineering, or a related field, or an equivalent combination of research and industry experience, with three or more years in machine learning, computer vision, or robotics.
• ML Pipeline: Hands-on experience designing, training, and tuning deep-learning models end to end in PyTorch.
• Method Development: Experience adapting, combining, or extending existing machine learning or vision methods for a new task, with changes that required understanding how those methods work internally.
• Computer Vision: Solid understanding of computer vision and 2D/3D geometry.
• Programming: Software development in Python, with working knowledge of C++.

Beneficial Skills

• Robot Control: Understanding of robot kinematics, feedback control, or visual servoing, or experience with high-precision manipulation. The team supports candidates in building this experience on the job.
• Model Deployment: Experience deploying optimized models with ONNX, TensorRT, or comparable frameworks, including on embedded or industrial GPU platforms.
• Robot Learning: Familiarity with reinforcement learning or imitation learning.
• Industrial Cameras: Experience with industrial 2D/3D cameras, lens models, camera calibration, or point clouds.
• Robotics Middleware: Experience with ROS or ROS 2.