Post Doc Res Assoc

Job Summary

The Kahlert School of Computing at the University of Utah invites applications for a 1-year post‑doctoral researcher for interdisciplinary work spanning machine learning, applied mathematics, computer science, and material sciences, with a heavy software development component. The successful candidate will perform research in the application of machine learning (ML) techniques to the finite element method (FEM) for composites and nonlinear materials, specifically dry fibers, and will work closely with the Air Force Research Laboratory (AFRL) and ARCTOS to integrate research into an existing JAX FEM framework jointly developed by the University of Utah, AFRL, and ARCTOS.

This position also carries an expectation that any candidate will spend 9 months of the academic year at the University of Utah under the supervision of Professors Varun Shankar and Robert M. Kirby, and 3 months on‑base at AFRL working with ARCTOS/AFRL while continuing collaboration with the University of Utah. Successful candidates will have the opportunity to continue working with ARCTOS as a research scientist to further this and other lines of research.

Salary Range: $60K – $70K – Post‑doc funding depending on experience, with additional benefits.

• Contribute to the development of a hybrid software framework for the finite element method and machine learning within the JAX Python library, including efficient implementations of classical numerical algorithms.
• Extend the hybrid FEA‑ML framework to include nonlinear cohesive zone models with simple traction‑separation laws in a modular manner such that additional laws can be incorporated in the future.
• Further extend the framework toward allowing discrete damage modeling via extended finite element methods, leveraging existing and actively developed AFRL libraries related to crack growth.
• Employ the hybrid FEM‑ML framework to demonstrate a machine‑learned crack model based on rich experimental data, using material information below the resolution of the mesh and the stress field as predicted by an FEA solution as inputs and providing predictions of crack initiation or propagation direction.
• Contribute to the hybrid FEA‑ML framework in the JAX Python library to support the simulation of dry‑fiber mechanics that incorporates the relevant physics and contact, as identified by AFRL. This includes implementing relevant algorithms and solvers for distributed GPU computing within the JAX Python library.

The Kahlert School of Computing is seeking a highly talented and committed individual with a demonstrated ability to work well with minimal supervision in a multi‑disciplinary research environment. Backgrounds in the engineering sciences, applied mathematics, physics, and computational sciences will be considered. Candidates comfortable with machine learning, the JAX library,