Differentiable Simulation Research Intern - Summer 2027
Core
Researching scalable differentiable algorithms (adjoint, automatic differentiation) to connect simulations with experimental data for calibration and data assimilation using AI/ML methods.
Role type
Research Intern (Differentiable Simulation)
Builds
GPU-native solvers, numerical algorithms, and reusable capabilities in CUDA-X libraries.
Domain
Scientific computing, computational engineering, AI/ML, GPU computing
Deliverable
production ML models
Required skills
PDEs, numerical linear algebra, iterative solvers, optimization, scientific computing, adjoint methods, automatic differentiation, inverse problems, data assimilation, Python, C++, parallel/GPU computing
Preferred skills
Multi-GPU/distributed execution, CUDA, profiling, benchmarking, model calibration, NVIDIA Warp, CUDA-X libraries, PyTorch, JAX, PETSc, Trilinos, MFEM, libCEED, OpenFOAM
Technologies
CUDA, CUDA-X, PyTorch, JAX, PETSc, Trilinos, MFEM, libCEED, OpenFOAM, NVIDIA Warp
Responsibilities
Prototype GPU-native solvers with adjoint and automatic differentiation techniques; Integrate AI-based methods with simulation algorithms for parameter calibration and data assimilation; Co-design numerical algorithms and differentiation strategies for GPU parallelism; Evaluate numerical accuracy, convergence, gradient quality, performance, and scalability; Identify promising methods and move them toward reusable capabilities in CUDA-X libraries
Seniority
Intern
