Computational Scientist 4

Pacific Northwest National Laboratory•Richland, WA

About The Position

The Atmospheric, Climate, and Earth Sciences Division is seeking a computer scientist to lead software engineering and development for the Energy Research and Forecasting (ERF) model, a GPU-enabled, compressible atmospheric flow solver built on the AMReX adaptive mesh refinement framework. The role involves developing performant, portable numerical algorithms for atmospheric dynamical cores and ensuring correctness and efficiency across CPU and GPU architectures. The scientist will lead GPU portability and optimization for heterogeneous DOE exascale and pre-exascale systems, and oversee deployment, build systems, continuous integration, and performance validation on DOE high-performance computing systems. The position also collaborates with the multi-institution ERF development team to extend model capabilities in areas such as terrain representation, urban simulation, land surface modeling, atmospheric chemistry, aerosols, and cloud microphysics. The scientist mentors junior software engineers and postdoctoral researchers; contributes to code review, documentation, and open-source software practices; authors technical and scientific publications and software releases; and presents ERF capabilities at DOE program reviews and scientific conferences. The broader role encompasses designing and developing computer and computational systems, advancing theories of computation and algorithms, and producing new theories and technical advances.

Requirements

BS/BA and 7 or more years of relevant experience, or MS/MA and 5 or more years of relevant experience, or PhD and 3 or more years of relevant experience.

Nice To Haves

PhD in Computational Science, Engineering, Applied Mathematics, Atmospheric Science, or a closely related field, plus demonstrated experience in scientific software development.
Hands-on experience and leadership developing production-quality scientific or engineering software in C++ for high-performance computing environments.
Experience porting and optimizing scientific codes for GPU architectures using CUDA, HIP, SYCL, or portable abstraction layers.
Experience deploying and validating large-scale simulation codes on DOE HPC systems.
Experience with the AMReX framework or comparable block-structured adaptive mesh refinement libraries.
Experience developing, extending, or deploying the ERF model.
Strong software engineering practices, including version control, testing and continuous integration, code review, and technical documentation.
Excellent written and verbal communication skills, with a record of technical publications and/or open-source software contributions.
Expertise in compressible dynamical core numerics for atmospheric or geophysical flow modeling.
Track record of peer-reviewed publications on atmospheric models.
Experience mentoring students or postdoctoral researchers and contributing to multi-institution DOE software projects.

Responsibilities

Lead software engineering and development efforts for the Energy Research and Forecasting (ERF) model.
Design, develop, and maintain performant, portable numerical algorithms for atmospheric dynamical cores, ensuring correctness and efficiency across CPU and GPU architectures.
Lead GPU hardware portability efforts, optimizing ERF for heterogeneous DOE exascale and pre-exascale systems, including NVIDIA, AMD, and Intel GPU platforms.
Oversee deployment, build systems, continuous integration, and performance validation of AMReX-based models on DOE high-performance computing systems and PNNL institutional computing resources.
Collaborate with the multi-institution ERF development team to extend model capabilities, including terrain representation, urban simulation, land surface modeling, atmospheric chemistry, aerosols, and cloud microphysics.
Mentor junior software engineers and postdoctoral researchers contributing to ERF and related AMReX-based codes.
Contribute to code review, documentation, and open-source software practices.
Author and co-author technical and scientific publications and software releases.
Present ERF capabilities at DOE program reviews and scientific conferences.
Design and develop computers and computational systems, including hardware and software, with a focus on architectures, languages, and applications.
Research and advance theories of computation, including algorithm design and efficiency and human interaction with computers and computing processes.
Develop work products involving new theories and technical advances in the field.