Graduate Software Engineer, Accelerator Modelling 2027

About this role

Location: Bristol, UK Start date: September 2027 Application deadline : Midnight, 29th November 2026 We encourage you to apply early, as we may close applications before the advertised deadline. The Graduate Programme at Graphcore Join Graphcore in a permanent role and start your career with two years of hands-on learning. From day one, you’ll be trusted to take ownership, share your ideas and help shape what we’re creating.

You’ll tackle challenging problems alongside a team who support your development and are as excited about your success as you are. What you’ll work on Put your C++ skills to work modelling future AI accelerators before they exist and help shape how Graphcore engineers design them. As a Graduate Software Engineer in Accelerator Modelling, you’ll develop, test and refine models and simulations for the next-generation AI accelerators.

Your work will help predict performance, power consumption and scalability, giving engineers evidence to compare approaches before hardware decisions are made. You’ll investigate how complex workloads behave, contribute to planning, design, implementation and testing, and help other teams integrate future hardware versions. You’ll also help scale distributed modelling environments. It’s a chance to see how software models connect architectural ideas with future hardware.

Your team and Support The Accelerator Modelling Team works across architecture, hardware and software. You'll be working in a small team, team lead and all senior members would support you with every step on your way. Modelling team faces a lot of the unknowns and challenging problems at the cutting edge of AI technology. There’s a buddy system in place, so you will always have a point of contact , ready to answer all questions or help with anything.

Decisions are driven by evidence from the models and clear engineering judgement. You’ll be encouraged to ask questions, challenge existing approaches and share ideas for making things better. How you’ll learn and grow During the two-year programme, you’ll build practical experience in C++ development, hardware simulation and performance modelling, building skills close to architecture, software and hardware design. Developing and debugging modelling code will help you understand how complex workloads behave and how to test whether a model captures that behaviour.