Job Details:
Job Description:
Scaling foundation models to the physical world is one of the hardest challenges in AI today. This role moves beyond simulation, which puts intelligence directly onto real hardware. You'll build the end-to-end stack for robotic intelligence as part of the Intel's Open Edge Platform and OpenVINO toolkit. The work is entirely open-source under the Apache 2.0 license to create the bridge between high-level reasoning and real-time physical action.
About the Role
This role spans the entire lifecycle of physical AI, from training large-scale Vision-Language-Action (VLA) models to building lightweight runtimes for real-time deployment. It's a position for engineers who work at the intersection of high-level research and low-level performance optimization. You'll ensure that complex policies don't just work in a paper, but run reliably on edge hardware with minimal latency.
What You'll Do
- Implement and fine-tune state-of-the-art Vision-Language-Action policies for robotic manipulation and control.
- Export and optimize VLA models for edge deployment without accuracy loss via export pipelines, graph optimization, and precision calibration to ensure policies run at high frequencies on constrained hardware.
- Develop safety runtimes to manage action clamping, velocity limits, and workspace bounds for reliable real-world operation.
- Build and maintain model-agnostic inference APIs that abstract across different robotic platforms and hardware backends.
- Support deployment across a range of edge accelerators and inference runtimes to ensure broad hardware compatibility.
- Write and publish research papers at top-tier venues to contribute to novel findings in vision, physical AI, model optimization, and robotic learning.
Qualifications:
What You Bring
- Extensive experience with PyTorch and PyTorch Lightning, including distributed training for large-scale models.
- Hands-on experience working with robotic systems, including industrial robots, EMR (autonomous mobile robots), and humanoid robots. Strong familiarity with robotics hardware and software stacks such as cameras and depth sensors, robot control interfaces, Robotics Operating Systems, and real-world robot deployment and integration.
- Deep understanding of Vision-Language-Action (VLA) architectures and imitation learning techniques applied to embodied agents and robotic systems.
- Technical expertise in model export and optimization, including graph compilation, operator fusion, and precision calibration for efficient edge and on-device inference.
- Familiarity with inference runtime backends such as OpenVINO, ONNX Runtime, and ExecuTorch.
- Ability to write clean, modular Python code and manage complex deployment dependencies across heterogeneous training, simulation, and production environments.
Nice to Have
- PhD in robotics, machine learning, computer vision, or a related field.
- Contributions to open-source robotics, embodied AI, or model optimization projects.
- Background in low-level performance profiling, memory optimization, and hardware-aware optimization for edge and robotic devices.
Why Join
- Build open-source tools that the entire robotics community can use and modify.
- Own the full pipeline, from training a model to watching it control a physical arm.
- Prioritize performance and real-world utility over theoretical cloud metrics.
- Contribute to a hardware-agnostic stack that supports the next generation of physical intelligence.
Job Type:
Experienced Hire
Shift:
Shift 1 (Netherlands)
Primary Location:
Virtual Netherlands
Additional Locations:
Business group:
The Client Computing Group (CCG) is responsible for driving business strategy and product development for Intel's PC products and platforms, spanning form factors such as notebooks, desktops, 2 in 1s, all in ones. Working with our partners across the industry, we intend to deliver purposeful computing experiences that unlock people's potential - allowing each person use our products to focus, create and connect in ways that matter most to them.
Posting Statement:
All qualified applicants will receive consideration for employment without regard to race, color, religion, religious creed, sex, national origin, ancestry, age, physical or mental disability, medical condition, genetic information, military and veteran status, marital status, pregnancy, gender, gender expression, gender identity, sexual orientation, or any other characteristic protected by local law, regulation, or ordinance.
Position of Trust
N/A
Work Model for this Role
This role is available as a fully home-based and generally would require you to attend Intel sites only occasionally based on business need. However, you must live and work from the country specified in the job posting, in which Intel has a legal presence. Due to legal regulations, remote work from any other country is unfortunately not permitted. * Job posting details (such as work model, location or time type) are subject to change.
*