PyTorch Engineer

About the job

Help make Graphcore hardware feel native inside the ML frameworks engineers use every day.

As a Software Engineer in our PyTorch team, you will help build the software that connects Graphcore accelerators with state-of-the-art machine learning frameworks. You will design, implement, optimise and support features across PyTorch, Triton and related technologies.

Your work will help ML engineers and researchers get more from Graphcore hardware, without changing how they want to build. That means better compatibility, stronger performance and cleaner developer experiences.

You will contribute to upstream PyTorch, support our hybrid Python and C++ extension, and help integrate compiler backend capabilities with torch.compile. You will learn fast, solve real engineering problems and build expertise in a large open-source ecosystem.

The team and culture

You will join the PyTorch team within Frameworks, working in a SCRUM team that ships features, reviews code, writes documentation and supports users. Work happens through clear ownership, practical decisions and close communication across software, AI/ML, research and hardware teams.

The team values maintainable code, thoughtful design and curiosity about unfamiliar systems. Engineers are trusted to investigate, speak up and improve the codebase as they learn.

This is a team for people who like open-source communities, complex technical problems and visible impact. You will move quickly, but with care for users and long-term quality.