Graduate Software Engineer, AI Compute Libraries & Performance 2027
Core
Develop C++ compute kernels and libraries for machine learning workloads on Graphcore hardware, focusing on performance optimization and correctness.
Role type
Graduate Software Engineer (AI Compute Libraries & Performance)
Builds
High-performance compute kernels (GEMM, convolutions, reductions) and regression test suites for Graphcore AI hardware.
Domain
AI hardware acceleration, numerical computing, low-level software engineering
Deliverable
production ML models
Required skills
C++11, linear algebra, numerical methods, profiling, microbenchmarking, regression testing, parallel programming
Preferred skills
Machine learning frameworks, code optimization, performance analysis
Technologies
C++, Python, Graphcore hardware
Responsibilities
Implement and improve compute kernels for matrix multiplication and convolutions; use profiling to measure execution speed and investigate performance bottlenecks; build regression tests and numerical checks to ensure library correctness.
Seniority
Graduate, 2-year learning programme
