Lead GPU Kernel Optimization Engineer
AMD · Hyderabad · 15+ yrs experience · Posted 2026-06-24
Tech stack: C, Python
Apply on the company site · Get a referral for this role
AMD salary & ratings · AMD interview process · More live openings
About the role
GPU Kernel Optimization Engineer
Responsibilities:
- GPU Kernel Optimization:
- Develop and optimize low-level GPU kernels to accelerate inference and training of large machine learning models. Maximize computational efficiency and reduce execution time while ensuring model accuracy.
- Multi-GPU and Multi-Node Optimization: Design and implement strategies for distributed model training and inference across multiple GPUs and nodes. Address data parallelism and model parallelism challenges to fully utilize available resources.
- Performance Profiling: Profile and analyze system and application performance to identify bottlenecks and areas for improvement. Use profiling tools to understand and optimize hardware resource utilization.
- Parallel Computing: Leverage parallel computing techniques to improve the scalability and performance of machine learning workloads. Implement multi-threading and GPU synchronization techniques.
- Benchmarking and Testing: Develop benchmarks and testing procedures to assess the performance and stability of optimized models and frameworks.
- Ensure that the solutions meet or exceed the defined performance criteria.
- Documentation: Create detailed documentation of optimizations, best practices, and implementation guidelines to facilitate knowledge sharing and maintainable code.
Qualifications:
- Over +15 years of architecting software:
- Optimizing GPU kernels in C++17, C++20, and C++23.
- Working with complex frameworks/libraries such as ( PyTorch, VLLM, CUTLASS, Kokkos, etc.)
- Practical working knowledge of GPU/CPU architectures.
- Scripting knowledge (ex. Python)
- Knowledge of software development processes.
- Good coding practices and design pattern identification, and implementation
- Agile processes and procedures
- Over +10 years of working with:
- Optimization in assembly
- Strong experience in low-level GPU kernel optimization.
- Proficiency in CUDA and GPU programming.
- Experience with distributed computing and multi-GPU environments.
- Proficiency in performance profiling and optimization tools.
- Excellent problem-solving skills and attention to detail.
- Strong communication and teamwork skills.
Qualifications
- Over +15 years of architecting software:
- Optimizing GPU kernels in C++17, C++20, and C++23.
- Working with complex frameworks/libraries such as ( PyTorch, VLLM, CUTLASS, Kokkos, etc.)
- Practical working knowledge of GPU/CPU architectures.
- Scripting knowledge (ex. Python)
- Knowledge of software development processes.
- Good coding practices and design pattern identification, and implementation
- Agile processes and procedures
- Over +10 years of working with:
- Optimization in assembly
- Strong experience in low-level GPU kernel optimization.
- Proficiency in CUDA and GPU programming.
- Experience with distributed computing and multi-GPU environments.
- Proficiency in performance profiling and optimization tools.
- Excellent problem-solving skills and attention to detail.
- Strong communication and teamwork skills.
Responsibilities
- GPU Kernel Optimization:
- Develop and optimize low-level GPU kernels to accelerate inference and training of large machine learning models.
- Maximize computational efficiency and reduce execution time while ensuring model accuracy.
- Multi-GPU and Multi-Node Optimization:
- Design and implement strategies for distributed model training and inference across multiple GPUs and nodes.
- Address data parallelism and model parallelism challenges to fully utilize available resources.
- Performance Profiling: Profile and analyze system and application performance to identify bottlenecks and areas for improvement.
- Use profiling tools to understand and optimize hardware resource utilization.
- Parallel Computing: Leverage parallel computing techniques to improve the scalability and performance of machine learning workloads.
- Implement multi-threading and GPU synchronization techniques.
- Benchmarking and Testing:
- Develop benchmarks and testing procedures to assess the performance and stability of optimized models and frameworks.
- Ensure that the solutions meet or exceed the defined performance criteria.
- Documentation: Create detailed documentation of optimizations, best practices, and implementation guidelines to facilitate knowledge sharing and maintainable code.