AI Software Engineer — ExecuTorch

Job Title

AI Software Engineer — ExecuTorch

Role Summary

Develop and optimize ExecuTorch, a PyTorch-compatible framework for running AI models efficiently on edge and resource-constrained devices. Join a team working across model optimization, compilers, runtimes and hardware-accelerated backends targeting Arm architectures.

The role spans high-level model transformations down to low-level runtime and hardware behaviour, with responsibility for improving latency, memory, cycle count and power efficiency.

Experience Level

Mid-level — experienced software engineer. No explicit years-of-experience requirement provided.

Responsibilities

Primary responsibilities include development, profiling, and optimisation across the software stack:

• Develop and optimise components of the ExecuTorch compiler, runtime and developer tooling (Python APIs, model transformations, C/C++ runtime code).
• Analyse and profile AI workloads to identify bottlenecks and improve latency, memory use, cycle count and power efficiency.
• Implement and tune backends for Mali GPUs, Cortex-A and Cortex-M CPUs, Ethos-U NPUs and other edge platforms.
• Work across high-level model transformations and low-level runtime/hardware behaviour to solve performance problems.
• Drive technical work from idea through delivery and collaborate with engineers across Arm and the PyTorch open-source community.
• Share knowledge and, where appropriate, mentor other engineers.

Requirements

Must-have technical skills and experience, plus useful additional experience:

• Must-have: Professional experience developing, debugging and maintaining production-quality software in C++ and Python.
• Must-have: Experience with compilers, runtimes, or performance optimisation and profiling tools.
• Must-have: Practical understanding of how software executes on hardware and how to improve efficiency (latency, memory, cycles, power).
• Must-have: Ability to identify technical problems, define practical approaches and deliver solutions; effective communication and teamwork skills.
• Nice-to-have: Experience with PyTorch, ExecuTorch or other ML frameworks; ahead-of-time compilers, graph transformation or model optimisation.
• Nice-to-have: Experience with embedded/mobile/edge systems, hardware-accelerator programming, Vulkan, open-source development, or technical leadership/mentoring.

Education Requirements

Not specified.

About the Company

Company: Arm

Headquarters: Cambridge, United Kingdom

ARM is a global leader in semiconductor and software design, driving innovation in computing technology. The company specializes in designing processors and systems that provide the essential building blocks for electronic devices. ARM's architecture is widely used in smartphones, servers, and IoT devices, and its collaborative culture fosters bold thinking, diversity, and high-impact benefits for its talented workforce.