← Back to roles

SW engineer - AI frameworks

Semidynamics Technology Services, S.L.

BarcelonaCompilersKernel / LinuxFirmwareFULL_TIMEPosted Sep 25, 2026
Apply on Semidynamics Technology Services, S.L.

Overview

Open roles
38
At this company
See all
Similar roles
9
Same title, 7 companies
Search
Added in 30 days
38
At this company, still open

Skills named in this posting

  • C
  • C++
  • Python
  • LLVM
  • GCC
  • CUDA
  • PyTorch / TensorFlow
What You'll Bring Required: • Strong experience with modern C++ • C programming experience • Working knowledge of Python • Experience with Linux environments • Multithreading experience • Software profiling and debugging (gdb) • Strong problem-solving skills and attention to detail • English proficiency at C1 level Nice to Have: • PyTorch dispatcher/ATen internals • vLLM • SGLang • CMake • LLVM/GCC • Hardware/software interfacing experience • Multi-GPU deployment • C/C++ and Python interoperability (pybind11) • ONNX Runtime execution provider (EP) registration • Open source contributions • CUDA/ROCm • AI kernel development • Bare-metal programming • AI application development • Workload analysis Why Semidynamics? • Join one of Europe's most promising deep-tech semiconductor scale-ups • Accelerated career development path • Hybrid model: 4 days/week at our Barcelona city-center office, 1 day WFH • 1 week per year of "work from anywhere in the world" • Competitive compensation package

RISC-V ecosystem

Semidynamics Technology Services, S.L. in the RISC-V ecosystem

RISC-V International: Strategic

Semidynamics is a supplier of high bandwidth RISC-V IP cores. | Semidynamics is a fabless semiconductor company based in Barcelona, focused on advancing high-performance AI inference through next-generation processor technology.

  • Semidynamics designs fully customisable RISC-V processor IP (64-bit cores with wide vector units) in Barcelona
  • Launched Cervell, a scalable RISC-V NPU for AI workloads, positioning itself as "AI with RISC-V"
  • Ships a RISC-V SDK with ONNX Runtime support so AI models run on its cores

Same employer

View all 38