HPC Infrastructure Software Engineer (Europe-wide)

A technology company developing advanced engineering software is expanding the team behind its next-generation computing platform.

This is a hands-on, technically demanding position for an experienced HPC engineer who understands how to extract exceptional performance from modern CPU and GPU architectures. You will build and accelerate the kernels, numerical operators and parallel infrastructure that power large-scale simulation, CAE and scientific computing applications.

The role combines low-level optimisation with wider architectural influence. You will work close to the hardware while helping shape the performance strategy of a growing computational platform.

What you will be working on

• Creating high-performance numerical components for heterogeneous CPU/GPU environments.
• Building and tuning computational kernels and numerical operators.
• Moving workloads efficiently between CPU and GPU architectures.
• Improving memory access, data layouts, cache behaviour and host-device transfers.
• Optimising GPU kernels using CUDA, HIP, SYCL or comparable technologies.
• Investigating occupancy, warp utilisation, memory coalescing and compute efficiency.
• Accelerating CPU code through multithreading, SIMD and vectorisation.
• Improving sparse and dense linear algebra operations.
• Developing acceleration technology for numerical solvers and scientific workloads.
• Profiling complex applications to identify processor, memory and communication bottlenecks.
• Designing reusable HPC libraries and software components.
• Supporting multi-GPU and distributed computing where needed.
• Collaborating with numerical solver, CAE, AI and software architecture specialists.
• Contributing to technical direction, engineering standards and long-term platform decisions.

What you will bring

• A degree in computer science, computational science, applied mathematics, physics, engineering or a related subject.
• Significant professional experience in HPC or performance-critical software engineering.
• Excellent C or C++ programming ability.
• Commercial experience with at least one GPU programming environment, such as CUDA, HIP/ROCm, SYCL/DPC++, oneAPI or OpenCL.
• A strong understanding of heterogeneous computing and modern CPU/GPU architecture.
• Deep knowledge of parallel programming.
• Practical understanding of memory hierarchy, bandwidth, caching and compute optimisation.
• Experience developing and accelerating performance-critical computational kernels.
• Strong profiling and performance-analysis skills.

Experience that would be particularly valuable

• GPU kernel fusion, warp-level programming, occupancy tuning and shared/global memory optimisation.
• SIMD, SIMT, AVX, AVX2, AVX-512, vectorisation or NUMA optimisation.
• OpenMP, MPI, OpenACC, multi-GPU systems or distributed parallel computing.
• cuBLAS, cuSPARSE, cuSOLVER, CUTLASS, MKL/oneMKL, rocBLAS or rocSPARSE.
• Sparse matrix computation and numerical linear algebra.
• SpMV, SpMM, GEMM, iterative solvers, preconditioners or finite element operators.
• NVIDIA Nsight Compute, Nsight Systems, Intel VTune, Intel Advisor or rocprof.
• Triton or the development of custom AI/ML operators.
• CAE, CFD, FEM, computational mechanics or scientific simulation software.

Your background

This position could suit someone who has developed HPC infrastructure, numerical libraries, solver technology or other performance-critical scientific software. Relevant experience may have been gained within an HPC or semiconductor business, a simulation-software company, a supercomputing centre, a scientific computing organisation or an advanced engineering R&D team.