Argonne’s Leadership Computing Facility (ALCF) and Mathematics and Computer Science Division (MCS) is looking for a Postdoctoral appointee working at the intersection of scientific machine learning SciML and large-scale simulation codes.

The Postdoctoral appointee will engage with projects that directly integrate machine learning (ML) components into partial differential equations (PDE)-based simulations, allowing ML algorithms to function as part of the simulation framework. This role involves coupling simulations with ML, enabling simultaneous training of these components within the simulation environment. The goal is to significantly accelerate traditional PDE-based simulations, particularly in fluid dynamics and turbulent flows.

As a Postdoc in ALCF/MCS, you will design ML kernels as local operators, embedding them within large-scale simulation codes. The work will also explore the use of advanced solvers like PETSc (www.petsc.org) for these developments, expanding their SciML applicability to a wider range of scientific applications. The incumbent will collaborate with scientists from multiple domains to enhance simulation speeds and accuracy through ML techniques.

Position Requirements

Required skills and qualifications:

• Ph.D. (completed within the last 0-5 years) in applied mathematics, physics, computer science, mechanical engineering, aerospace engineering, or similar fields.
• Experience contributing to scientific software in C, C++, Python, or comparable languages.
• Strong background in computational fluid dynamics (CFD), machine learning (ML), and high-performance computing (HPC).
• Collaborative skills include working well with other scientists, divisions, laboratories, and universities.
• Effective oral and written communication skills with all levels of the organization.
• Ability to model Argonne’s core values: Impact, Safety, Respect, Integrity, and Teamwork.

Preferred skills and qualifications:

• Demonstrated experience with parallel algorithms, for example, using frameworks like MPI, OpenMP, CUDA, Kokkos, SYCL, or similar.
• Demonstrated experience running applications at scale (> 50 computing nodes). This could be AI training, simulation, or data analytics.
• Experience with PETSc (or other scientific libraries used as external solvers) and its integration with ML frameworks such as PyTorch.
• Knowledge of subgrid-scale models and closure models.
• Experience with computing gradients for PDE-constrained optimization problems, adjoint calculations and backpropagation.
• Ability to address dynamic stability issues of hybrid systems during training and operational phases.

Job Family

Job Profile

Worker Type

Time Type