AI Research and Development Engineer (Physical AI)

Location: Fully home-based in the Netherlands; remote work from other countries is not permitted. Occasional attendance at Intel sites may be required based on business needs.

Annual salary: €74,990–€139,760.

Company

Build open-source physical AI and robotics tools through the Intel Open Edge Platform and OpenVINO toolkit under the Apache 2.0 license.

What you will do

• Implement and fine-tune Vision-Language-Action policies for robotic manipulation and control.
• Export and optimize VLA models for real-time edge deployment using graph optimization, operator fusion, and precision calibration.
• Develop safety runtimes for action clamping, velocity limits, and workspace bounds.
• Build model-agnostic inference APIs across robotic platforms and hardware backends.
• Support deployment on edge accelerators and inference runtimes including OpenVINO, ONNX Runtime, and ExecuTorch.
• Write and publish research papers on vision, physical AI, model optimization, and robotic learning.

Requirements

• Extensive experience with PyTorch and PyTorch Lightning, including distributed training for large-scale models.
• Hands-on experience with industrial robots, autonomous mobile robots, humanoid robots, sensors, robot control interfaces, and real-world deployment.
• Strong understanding of Vision-Language-Action architectures and imitation learning for embodied agents.
• Expertise in model export, graph compilation, operator fusion, precision calibration, and efficient edge inference.
• Experience writing modular Python code and managing deployment dependencies across training, simulation, and production environments.
• Must live and work from the Netherlands.

Nice to have

• PhD in robotics, machine learning, computer vision, or a related field.
• Open-source contributions in robotics, embodied AI, or model optimization.
• Experience with performance profiling, memory optimization, and hardware-aware optimization.

Culture & Benefits

• Work on open-source tools available under the Apache 2.0 license.
• Own the full pipeline from model training to physical robot control.
• Prioritize real-world performance and utility over theoretical cloud metrics.
• Contribute to a hardware-agnostic stack for physical intelligence.
• Fully home-based work with occasional site attendance based on business needs.