AI Engineer

Position: Staff AI Engineer
Arm is building the foundational platform software for the next generation of Physical AI systems on Arm: software that lets AI applications and agents perceive, reason and act through real devices, reliably and on-device, across robotics, industrial automation and the smart home. The work sits at the intersection of distributed systems, edge computing and AI infrastructure. This is a rare opportunity to join a new team in Seattle and build new products and reference implementations that enable developers to put the latest Arm technology to work in their solutions.

Job Overview:

Building intelligent systems that interact with the physical world requires bringing together AI models, diverse hardware, and software frameworks across the cloud and edge. Developers need simpler ways to connect these technologies, orchestrate complex workflows, and build reliable applications that can operate across different devices and environments. Our team is building the open-source foundations and developer tools to make Physical AI more accessible, scalable, and reliable.

We are working to simplify how AI agents interact with physical devices, coordinate tasks, and operate across heterogeneous hardware environments. By bringing together AI, hardware, and developer ecosystems, we aim to make building and deploying real-world agentic applications as seamless as building software today. As a Staff AI Engineer, you will develop the tools and technologies that bring this vision to life.

You will design and build SDKs, frameworks, and reference applications that enable AI agents to interact with physical devices, orchestrate complex workflows, and operate across cloud and edge environments. You will integrate modern AI models, inference runtimes, and agent frameworks to create end-to-end applications, while developing evaluation methodologies to measure their reliability, performance, and effectiveness in real-world scenarios. You will work hands-on with real physical hardware, and edge computing platforms, collaborating with engineers, product teams, and industry partners to shape the future of Physical AI on Arm.

Responsibilities:

Design and implement features in our SDKs and reference services, on both the device side and the agent side Build reference applications and demos that connect agents to real devices across edge and cloud, using PyTorch-ecosystem inference runtimes and popular agent frameworks Develop evaluation harnesses that measure the reliability, latency and correctness of agent-to-device interactions, and use the results to improve the product Bring up and maintain a fleet of development hardware (single-board computers, cameras, sensors, robot platforms, edge servers) used for development, testing and demonstrations Contribute to the security and operational posture of the platform: identity, authorization, observability and safe failure behavior Write clear documentation, samples and learning material for external developers, and support the open-source community through issues, reviews and contributions Collaborate with architects, product managers and partner engineers on requirements, API design and release quality Required

Skills and Experience:

Professional software engineering experience building distributed systems, with strong Python and working proficiency in at least one of C++, Rust or Go Experience building and shipping AI-powered applications, including LLM tool use or agent frameworks, and integrating them with real systems Practical experience deploying ML models for inference, on servers or on edge devices, with PyTorch or a comparable framework Solid grounding in messaging, RPC, concurrency, failure handling and testing of networked software Comfortable with Linux, embedded single-board computers, containers and CI/CD workflows Clear written and verbal communication, and the ability to work across hardware, systems and ML disciplines Degree or equivalent experience in Computer Science, Electrical Engineering, Robotics or a related field'Nice to Have'

Skills and Experience:

Experience with publish/subscribe messaging systems or IoT protocols Familiarity with the Model Context Protocol or similar agent-tool interfaces Robotics or physical-AI background: ROS 2, robot arms, perception pipelines, or control loops On-device inference experience with Execu Torch, ONNX Runtime or vendor NPU tool chains, and LLM serving with vLLM or similar Experience designing evaluations or benchmarks for AI agents Contributions…