134K–252K USD / year
Machine Learning Scientist - Molecular Force Fields
Machine learningRemote — United States
Published on 2026-10-26
These details were extracted automatically from the original listing. They may not be complete or fully up to date — it's worth checking the original job post.
About this role
Develop and deploy machine-learned force fields to support the discovery of PFAS-lean or PFAS-free semiconductor materials. The role spans first-principles data generation, model development and validation, and production simulation campaigns, with an emphasis on communicating model uncertainty and limits.
About the company
SandboxAQ leverages artificial intelligence and quantum technology to improve decision-making processes across various sectors, including life sciences and medical devices. The company focuses on developing innovative solutions that address complex challenges in data analysis, research, and product management.
The team
PFAS team within the Chemical Simulation (ChemSim) group.
Stack
PythonPyTorchJAXMACENequIPAllegroTensorNetDeePMD
What you'll do
- Design and run high-throughput quantum chemistry and molecular dynamics campaigns to generate training data.
- Build reproducible, restartable workflows across cloud and HPC, and curate and version scientific datasets.
- Train, validate, and deploy machine-learned interatomic potentials for PFAS substitution chemistry.
- Develop and benchmark MLFF-based protocols for properties including pKa, diffusivity, viscosity, and thermal conductivity.
- Extend potentials to polymerization and plasma surface-etch chemistry in condensed-phase systems.
- Run production prediction campaigns and publish models, datasets, and documented pipelines for internal use.
- Report model errors and domain-of-validity limits, and collaborate with modeling, simulation, and materials teams.
What we're looking for
- PhD in computational chemistry, materials science, physics, machine learning, or a related field, with deep expertise in ML interatomic potentials or atomistic simulation.
- Have trained, validated, and deployed at least one machine-learned force field used by others.
- Experience generating first-principles training data with production quantum chemistry or electronic-structure codes.
- Fluency with descriptor-based, equivariant, and message-passing potential architectures, including their limitations.
- Strong Python skills and experience with PyTorch or JAX and GPU-cluster training.
- Experience integrating learned potentials with molecular dynamics engines and extracting converged observables.
- Hands-on classical molecular dynamics experience, including force-field selection, parameterization, system setup, and equilibration.
- Experience with automated computational workflows, scientific data provenance and versioning, model validation, cloud, and HPC.
Nice to have
- Experience with fluorine-containing chemistry, fluoropolymers, or PFAS-adjacent systems.
- Experience computing free energies with thermodynamic cycles or alchemical methods, or transport properties with Green-Kubo or non-equilibrium MD.
- Experience with long-range electrostatics, polarization, or implicit-solvation methods in learned potentials.
- Familiarity with generative molecular design frameworks.
Benefits
- Equity and performance-based incentives.
- Health, dental, and vision insurance.
- 401(k) with company match.
- Generous parental leave.
- Generous paid time off and flexible work arrangements.
- Dedicated learning budgets.