Software Engineer - GPU Performance

About the role

We are looking for a Software Engineer specialized in GPU performance to join CynLr’s visual intelligence team. You will translate the company’s vision‑algorithm and learning software stack into highly optimized GPU code, enabling robots to see and manipulate any object in real‑time.

Key responsibilities

• Implement low‑level CUDA kernels and integrate them into the vision pipeline.
• Perform detailed GPU performance analysis using NVIDIA profiling tools and optimise both kernel and system‑level execution.
• Design and maintain a pipelined image‑processing framework that balances workloads across CPU and GPU.
• Develop dynamic load‑balancing and runtime control‑flow mechanisms between kernels.
• Utilise NvidiaDirect to access peripheral memory (PCIe, USB, display) directly from the GPU.
• Create visual debugging tools for GPU memory without CPU transfer.
• Collaborate with algorithm, hardware and software teams to ensure seamless integration.

Required profile

• Strong background in C++/C and low‑level CUDA API development.
• Deep understanding of GPU hardware architecture and performance optimisation techniques.
• Experience with GPU profiling (NVIDIA Visual Profiler, Nsight Compute) and debugging.
• Proven ability to optimise mathematical models and neural‑network kernels for time‑weighted execution.
• Familiarity with state‑machine architecture, realtime computing and memory optimisation (MIMD, SIMD).

Required skills

• CUDA
• C++ / C
• Visual Studio toolchain
• NVIDIA Nsight Compute & Visual Profiler
• GPU kernel optimisation
• NvidiaDirect memory access
• Neural network kernel optimisation
• Omniverse (plus)
• CPU architectures: x86, x64, ARM
• Driver and low‑level OS development (BIOS, UEFI, bootloader)
• UI‑based application deployment