Senior Applied Research Scientist – GPU Native Numerical Algorithms

This is a fully remote position, open to applicants in Germany.

📋 Description

• Design and refine numerical algorithms tailored for contemporary NVIDIA GPU architectures, encompassing both implicit and explicit engineering simulations.

• Create linear and nonlinear solver methodologies, which include Newton-Krylov methods, multigrid and algebraic multigrid (AMG) techniques, domain decomposition, matrix-free algorithms, mixed precision, as well as sparse iterative and direct methods, along with preconditioning strategies.

• Explore GPU-native alternatives to traditional CPU-centric numerical methods, such as synchronization-avoiding Krylov methods, GPU-native multigrid and domain decomposition approaches, matrix-free implicit techniques, mixed-precision algorithms, and hybrid sparse direct/iterative methods.

• Assess algorithms across various domains including mechanics, contact mechanics, thermal-fluid systems, electromagnetics, semiconductor process and device simulations, electronic design automation (EDA), multiphysics, and related computer-aided engineering (CAE) tasks.

• Develop research prototypes, evaluate different approaches on representative workloads, and assist in transitioning successful concepts into NVIDIA software.

• Partner with teams from CUDA-X, Warp, solver engineering, NVIDIA Research, as well as academic institutions and industry collaborators.

• Define the long-term applied research agenda for GPU-native numerical techniques and their progression towards AI-native computational engineering.

⛳️ Requirements

• A PhD or equivalent experience in computational mechanics, applied mathematics, scientific computing, computer science, aerospace, mechanical or civil engineering, or a similar technical discipline.

• Over 5 years of pertinent work or research experience.

• Background in research or engineering involving PDE discretization, finite element methods, finite volume techniques, discontinuous Galerkin methods, nonlinear solvers, sparse linear algebra, preconditioning, or high-performance computing.

• Proficiency in developing numerical software using C++ and Python.

• Experience in developing or optimizing CUDA or GPU programming.

• Familiarity with profiling, benchmarking, numerical validation, or performance analysis to enhance algorithms on GPU or multi-GPU platforms.

• Ability to articulate technical trade-offs effectively and collaborate across research, engineering, product, and partner teams.

• Experience in implicit structural dynamics, nonlinear mechanics, contact problems, computational fluid dynamics (CFD), electromagnetics, multiphysics, semiconductor simulations, EDA, CAE, or CAD-integrated engineering processes.

• Contributions to or hands-on experience with PETSc, Trilinos, MFEM, libCEED, OpenFOAM, NVIDIA Warp, CUDA-X, cuSPARSE, cuSOLVER, or similar computational science frameworks.

• Experience within industrial simulation, EDA, semiconductor, CAE, or CAD ecosystems, including tools like Ansys, Abaqus, LS-DYNA, Siemens Simcenter, Dassault SIMULIA, Altair, Cadence, Synopsys, COMSOL, MathWorks, or equivalent in-house solver and design platforms.

• Familiarity with distributed solvers utilizing MPI, NCCL, asynchronous methods, or performance analysis on GPU clusters.

• Published works, patents, contributions to open-source projects, or deployed software in computational science forums or communities such as SC, SIAM CSE, SIAM SISC, CMAME, IJNME, JCP, AIAA, USNCCM, WCCM, or related fields.

🏝️ Benefits

• Access to NVIDIA software, systems, and research platforms dedicated to accelerated computing.

• Opportunity to engage with CUDA-X, Warp, NVIDIA Research, academic institutions, and industry partners.

• Potential to convert research prototypes into NVIDIA software functionalities.

• Collaborative environment across research, engineering, product, and partner teams.