Graduate Software Engineer, AI Compute Libraries & Performance 2027

Description

Permanent role on a two-year graduate programme, starting September 2027.

Applications close at midnight on 29th November 2026 and may close earlier, so apply soon.

You will help AI models run efficiently on the company's next-generation hardware, working on C++ libraries and compute kernels for machine learning in the AI Compute Libraries and Performance team.

The company builds AI compute across semiconductors, software and datacenter-scale systems. It is owned by SoftBank Group, and you will work alongside AI researchers, chip designers and software engineers.

The team builds high-performance compute libraries that help customers get more from the hardware, combining machine learning, numerical computing and accelerator performance.

You start on clearly scoped projects with guidance from engineers in performance and low-level software. A buddy helps you settle in, and your manager meets you regularly to set goals, give feedback and plan time for learning.

The team uses profiling, experiments and technical reviews to compare approaches and weigh trade-offs. Questions and alternative ideas are welcome.

Over two years you build performance engineering skills on kernels, benchmarks and tests, learning where time goes and how to keep optimisations numerically correct.

You will hear the outcome by early December at the latest. Successful candidates complete a technical assessment, and final-stage candidates attend in-person assessments in early spring.

Responsibilities

• Implement and improve kernels for operations such as GEMM, convolutions, reductions, elementwise operations and fused kernels that combine several operations.
• Profile code and run microbenchmarks to measure execution speed, and study how threading, cache locality, memory layout and kernel launch efficiency affect performance.
• Write regression tests and numerical checks, investigate defects, and raise the quality of the libraries customers rely on.

Requirements

• Bachelor's or Master's degree in computer science, maths, machine learning, data science or a related field.
• Experience with C++11.
• Familiarity with Python or scripting tools for automation and testing.
• Understanding of linear algebra, numerical methods or scientific computing, plus problem solving and a collaborative approach to unfamiliar technical challenges.
• Show these skills through coursework, a dissertation, personal or team projects, research or other experience: what you built or investigated, how you approached it and what you learned.
• Additional strength, not essential: coursework or experience in machine learning frameworks, parallel programming, code optimisation, maths libraries or performance analysis.

Conditions

Permanent contract.

You must already have the right to work in the UK for the full duration of the programme. Visa sponsorship is not available for early careers roles.

The company is an equal opportunity employer and will support interview adjustments on request.

Benefits

• Flexible working to balance work and personal life.
• Pension with up to 5% matched.
• Phantom equity, so you share in the company's success.
• Office spaces designed for collaboration, with free food and an on-site barista.
• Income protection and life assurance for you and your loved ones.
• Electric vehicle lease through salary sacrifice.
• Flexible benefits: pick extra options such as private medical insurance and dental cover.
• Optional benefits: dental cover, health cash plan, private medical insurance, cycle to work scheme and give as you earn.