Senior Architect- Molecular Dynamics

NVIDIA•Santa Clara, CA

About The Position

NVIDIA is looking for an architect with extensive experience in molecular dynamics to help us push the boundaries of scaling laws in physical, chemical, and biological sciences. You will work with both internal and external software developers, research scientists, and users to build broadly applicable primitives for the current and next generation of molecular dynamics engines. If you're driven, collaborative, and eager to make an impact at scale, we want to hear from you! What you'll be doing: Developing and deploying functional forms for modern forcefields. This includes implicit solvent, explicit solvent, and polarizable models. These forms are enabled for both classical and machine-learned interatomic potentials. Prototyping and implementing enhanced sampling methods, including but not limited to: Gibbs sampling, umbrella sampling, REST2, SAMS, lambda dynamics, r-RESPA, and machine-learned samplers. Integrating our libraries with various FEP, MD, and forcefield fitting workflows. Some relevant examples are: relative and absolute binding free energy calculations, equilibrium and non-equilibrium protocols, fitting forcefields in the context of both ab initio QM data and condensed phase data. Working with our CUDA-X library team and other engineering teams to build GPU-friendly algorithms to ultimately accelerate the hardest and most time-consuming kernels in MD and making them widely accessible to developers around the world.

Requirements

A Master’s or Ph.D. in Computer Science, Chemistry, Physics, or a related field (or equivalent experience).
8+ years of experience of methods development in MD, forcefield parameterization, and/or free energy methods.
Proficiency in Python, C++, and/or CUDA, ideally in a setting with established best-practices such as code-reviews, unit testing, integration testing, etc.
A deep understanding of trade-offs between practicality, generalizability, and performance of the solutions, and articulating them to both technical and non-technical experts.

Nice To Haves

Meaningful contributions to a major MD engine. (e.g. OpenMM, GROMACS, AMBER, etc.)
Experience with ML frameworks such as jax, pytorch, or tensorflow and the machinery behind automatic differentiation.
Numerical analysis and methods development (e.g. quantifying error propagation, mixed precision and fixed precision modes, bitwise determinism, etc.).
Track record of starting and landing initiatives that span engineering, product, and/or research, etc.

Responsibilities

Developing and deploying functional forms for modern forcefields.
Prototyping and implementing enhanced sampling methods
Integrating our libraries with various FEP, MD, and forcefield fitting workflows.
Working with our CUDA-X library team and other engineering teams to build GPU-friendly algorithms to ultimately accelerate the hardest and most time-consuming kernels in MD and making them widely accessible to developers around the world.