Software Engineer Summer Intern AI Compute Libraries and Performance 2027
Graphcore is a SoftBank-owned AI-compute company developing AI processors, systems, and software for machine-learning workloads.
Funding history
About Graphcore
Graphcore develops Intelligence Processing Units (IPUs) and the Poplar SDK for building and running machine-learning applications. It continues operating under the Graphcore name as a wholly owned SoftBank Group subsidiary.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will contribute to C++ libraries and compute kernels for machine-learning workloads. You will implement and improve operations such as matrix multiplication, convolutions, reductions, elementwise operations, and fused kernels. You will profile and benchmark code, investigate performance factors, build regression and numerical tests, and resolve library issues.
Requirements
- Currently studying towards a bachelor’s or master’s degree in Computer Science, Maths, Machine Learning, Data Science, or a related field
- Experience with C or C++11
- Familiarity with Python or scripting tools for automation and testing
- Understanding of linear algebra, numerical methods, or scientific computing
- Problem-solving skills
- Collaborative approach to unfamiliar technical challenges
- Right to work in the UK for the full duration without visa sponsorship
Responsibilities
- Develop and improve compute kernels
- Implement matrix multiplication, convolutions, reductions, elementwise operations, and fused kernels
- Profile and benchmark execution performance
- Investigate threading, cache locality, memory layout, and kernel-launch efficiency
- Build regression tests and numerical checks
- Investigate issues and improve library quality
Benefits
- Flexible working
- 5 days of annual leave
- Free food
- Ice cream
- On-site barista
- Access to O’Reilly online learning platform
Hiring Process
Submit application questions, CV, and cover letter; successful candidates complete a technical assessment; final-stage candidates attend in-person assessments.