Post Doctorate RA - HPC Proxy application-driven Codesign
The Data Sciences & Machine Intelligence group in the Advanced Computing, Mathematics, and Data Division at PNNL seeks a dynamic Post-Doctoral Data Scientist to join the group to lead and support codesign activities by analyzing real-world compute-intensive and data-intensive applications on heterogeneous computing platforms. This is an excellent opportunity to hone and develop your scientific career in an outstanding research institution by joining an interdisciplinary research team that focuses on key topics spanning data sciences and machine intelligence to advance scientific discovery. The primary emphasis of this position will be growing existing and crafting new capabilities in broad areas of:
(1) Analysis, modeling, and assessment of existing and emerging class of converged applications which can potentially consolidate Graph Analytics, Machine Learning/Artificial Intelligence and classic HPC simulations.
(2) Research and develop new algorithms and implementations targeting heterogeneous platforms.
(3) Collaborate closely with subject matter experts to solve challenging domain problems
(4) Lead and contribute to the publication and presentation of results in high impact scientific computing journals and conferences, and sponsoring agencies.
(5) Mentor and train graduate and undergraduate interns.
In addition to performance analysis and profiling on HPC platforms, knowledge of the latest advancements in common HPC parallel programming models such as OpenMP, MPI, CUDA, etc. are very relevant. Experience in developing parallel ML/AI workloads will be preferred.
Proficiency in parallel computing concepts, basic CS algorithms, HPC hardware architecture for performance optimization, , fundamentals of Graph Analytics and Machine Learning/Artificial Intelligence, experience in GPGPU programming (such as OpenMP offload/CUDA/HIP/OpenCL).
Proactive, highly motivated self-starter with demonstrated experience with contributing and leading tasks on projects with multi-disciplinary teams.
Proficiency in performance profiling/analysis/modeling of codes developed using HPC programming models and runtimes.
Demonstrated ability to develop approaches and solutions to complex problems in the forms of proposals, software, documents or other work products.
Candidates must have received a PhD within the past five years (60 months) or within the next 8 months from an accredited college or university.
Ph.D. degree in Computer Science, Electrical Engineering, Applied Mathematics, or related field.
Peer-reviewed publication record in Data Sciences, Machine Learning or a closely related area.