AI-Driven Numerical Methods Scientist

NVIDIA; New York (NY); On-site; USD 192,000 - 357,000; Full time

Benefits offered by this job

Equity

Benefits

Job summary

NVIDIA is seeking an applied researcher to fuse machine learning with numerical algorithms, accelerating solvers on modern GPUs. The role focuses on AI-guided multigrid, differentiable simulation, and hybrid numerical/ML methods across NVIDIA solver and simulation platforms.

PhD or equivalent with 5+ years in ML and scientific computing is required. Experience with PyTorch/JAX, Python, PDEs, and preconditioning is essential. Equity and benefits accompany this role at NVIDIA Research.

Qualifications

• PhD or equivalent in CS/ML/scientific computing or related field.
• Minimum 5 years of work/research experience.
• Strong background in ML and scientific computing linking ML with numerical algorithms.
• Experience with PyTorch/JAX, Python, and GPU computing.
• Knowledge of PDEs, sparse linear algebra, iterative solvers, and preconditioning.

Responsibilities

• Research AI-assisted numerical methods to improve convergence and speed of large-scale simulations.
• Invent learned coarse spaces, learned preconditioners, AI-guided multigrid, and solver-control policies.
• Build solver-in-the-loop pipelines with meshes, operators, and physics constraints.
• Define evaluation metrics for numerical impact (convergence, stability, speedup).
• Collaborate across NVIDIA Research and product/engineering teams.

Skills

Machine Learning

Scientific Computing

Python

GPU Computing

PDEs

Sparse Linear Algebra

Iterative Solvers

Preconditioning

Differentiable Programming

Optimization

Education

PhD or equivalent

Tools

PyTorch

JAX

CUDA

Job description

NVIDIA is seeking an applied researcher to fuse machine learning with numerical algorithms, accelerating solvers on modern GPUs. The role focuses on AI-guided multigrid, differentiable simulation, and hybrid numerical/ML methods across NVIDIA solver and simulation platforms.

PhD or equivalent with 5+ years in ML and scientific computing is required. Experience with PyTorch/JAX, Python, PDEs, and preconditioning is essential. Equity and benefits accompany this role at NVIDIA Research.