Senior HPC / CUDA Engineer

EPAM Systems; United States; Remote; USD 140,000 - 210,000; Full time

Job summary

EPAM Systems in the United States is seeking a Senior HPC / CUDA Engineer to optimize seismic imaging and Full Waveform Inversion workloads on modern GPU platforms. You will design CUDA kernels, implement high-performance C++/Python components, and collaborate with experts across HPC, geophysics, and software engineering.

The role emphasizes performance profiling, multi-GPU scalability, MPI/NCCL/NVSHMEM, and robust, portable scientific software.

Qualifications

• Strong HPC and GPU software development experience.
• Advanced CUDA programming and kernel optimization skills.
• Experience with distributed multi-GPU computing and parallel programming.

Responsibilities

• Design, develop, and optimize CUDA kernels and GPU-accelerated applications for HPC workloads.
• Analyze and improve application performance across NVIDIA GPU platforms.
• Profile and identify bottlenecks at kernel, PTX, memory, and multi-GPU levels.
• Optimize seismic imaging and FWI workloads for efficiency and scalability.
• Develop distributed GPU solutions using MPI, NCCL, and NVSHMEM.
• Apply knowledge of GPU architecture and memory hierarchy to improve performance.
• Develop and maintain high-quality C++ and Python components.
• Debug runtime issues to ensure reliability of performance-critical code.
• Contribute to seismic data preprocessing, post-processing, visualization, and analysis tools.
• Work with data formats like HDF5 and collaborate with domain experts.

Skills

HPC

CUDA

C++

Python

GPU performance

Multi-GPU

MPI/NCCL/NVSHMEM

Seismic computing

Performance analysis

Tools

MPI

NCCL

NVSHMEM

HDF5

Job description

We are looking for a Senior HPC / CUDA Engineer to optimize high-performance scientific computing applications running on modern GPU platforms. You will work on computationally intensive workloads in the geoscience domain, focusing on seismic imaging and Full Waveform Inversion (FWI). The role combines low-level CUDA optimization, modern C++ development, GPU performance analysis, and distributed multi-GPU computing to improve the performance, scalability, and efficiency of scientific applications. You will collaborate with experienced engineers working across HPC, numerical methods, GPU computing, and seismic data processing, contributing to the evolution of performance-critical software and complex computational workflows.

Responsibilities

• Design, develop, and optimize CUDA kernels and GPU-accelerated applications for computationally intensive workloads
• Analyze and improve application performance across consumer and data-center NVIDIA GPU platforms
• Profile and investigate performance bottlenecks at the kernel, PTX, hardware, memory, and multi-GPU communication levels
• Optimize seismic imaging and Full Waveform Inversion (FWI) workloads to improve computational efficiency and scalability
• Develop and maintain distributed and multi-GPU computing solutions using technologies such as MPI, NCCL, and NVSHMEM
• Apply knowledge of GPU architecture, execution models, memory hierarchy, and modern CUDA capabilities to improve performance
• Develop and maintain high-quality scientific software components using C++ and Python
• Debug complex runtime issues and ensure the reliability and maintainability of performance-critical applications
• Contribute to the development and maintenance of scientific tools for seismic data preprocessing, post-processing, visualization, and analysis
• Work with scientific data formats and storage technologies, including HDF5
• Collaborate with domain experts and engineering teams to analyze computational requirements, evaluate optimization approaches, and deliver effective technical solutions

Requirements

• Strong experience in high-performance computing (HPC) and GPU-accelerated software development
• Advanced knowledge of CUDA programming and GPU performance optimization
• Deep understanding of NVIDIA GPU architecture, GPU execution models, memory hierarchy, and kernel optimization techniques
• Experience with GPU profiling, performance analysis, and systematic identification of computational bottlenecks
• Strong C++ development skills and experience working with performance-critical applications
• Practical experience optimizing computationally intensive workloads on modern GPU platforms
• Understanding of low-level GPU execution and performance analysis, including PTX
• Experience with distributed or multi-GPU computing and parallel programming concepts
• Strong analytical and problem-solving skills, with the ability to investigate complex performance and runtime issues
• Experience working with large-scale computational workloads where performance, scalability, and reliability are essential

Nice to have

• Experience with seismic imaging, Full Waveform Inversion (FWI), or other geophysical computing applications
• Knowledge of numerical methods and computational algorithms used in scientific computing
• Hands‑on experience with MPI, NCCL, and NVSHMEM
• Experience with HDF5 and scientific data processing or visualization tools
• Familiarity with seismic data preprocessing and post-processing workflows
• Experience developing or maintaining geoscience software and scientific computing tools
• Knowledge of modern CUDA features and advanced GPU optimization techniques
• Experience with build systems, software deployment, and debugging workflows for HPC applications
• Previous experience in other performance‑critical domains, such as game engines, simulation, or scientific computing