Key skills
GPU programmingC/C++ programmingCUDAHIPROCmGPU profiling toolsDebugging skillsLinux experiencePerformance tuningC++C
About this role
AMD is a leading company focused on building innovative products that enhance next-generation computing experiences. They are seeking a GPU application engineer to enable AMD GPU acceleration through HIP and to optimize applications that currently use CUDA backends.
Responsibilities:
- Port CUDA-based GPU code to HIP, including kernels, runtime logic, and build systems
- Design and implement HIP backends for applications without existing AMD GPU support
- Enable and integrate HIP ray tracing (HIPRT) where applicable
- Optimize GPU kernels and application pipelines for AMD GPU architectures
- Debug correctness, performance, and stability issues in application-level GPU code
- Work with profiling tools to identify and resolve bottlenecks
- Collaborate with internal compiler/runtime teams to escalate lower-level issues when needed
- Maintain clean, portable GPU code across vendors
Requirements:
- Hands-on experience developing and optimizing mature CUDA-based applications
- Ability to port CUDA-based GPU code to HIP, including kernels, runtime logic, and build systems
- Design and implement HIP backends for applications without existing AMD GPU support
- Enable and integrate HIP ray tracing (HIPRT) where applicable
- Optimize GPU kernels and application pipelines for AMD GPU architectures
- Debug correctness, performance, and stability issues in application-level GPU code
- Work with profiling tools to identify and resolve bottlenecks
- Collaborate with internal compiler/runtime teams to escalate lower-level issues when needed
- Maintain clean, portable GPU code across vendors
- Bachelor's or Master's degree in Computer Science, Computer Engineering, Electrical Engineering, or equivalent
- Advanced software development experience, with significant experience in GPU programming
- Strong C/C++ programming skills
- Hands-on experience with CUDA (kernel development, memory management, streams/events)
- Experience developing GPU-accelerated applications
- Solid understanding of GPU programming fundamentals
- Strong debugging skills for GPU kernels and application pipelines
- Experience developing on Linux
- Experience with HIP and ROCm
- Experience porting CUDA code to HIP (manual or hipify-based)
- Familiarity with HIPRT or GPU ray tracing concepts
- Experience with GPU profiling tools
- Experience working in large, multi-backend codebases
- Familiarity with NVIDIA OptiX or similar GPU ray tracing APIs