SciML Computational Scientist — Fusion Plasma & Real-Time Control

Fusionality; Vaud; Hybrid; CHF 120.000 - 180.000; Vollzeit

Benefits dieser Stelle

Immigration assistance

Equity compensation

Competitive salary

Relocation assistance

Zusammenfassung

Fusionality in Lausanne is seeking a Computational Scientist specialized in Scientific Machine Learning (SciML) to design differentiable solvers for fusion plasma problems. You will work close with control and physics teams to build end-to-end, differentiable solvers in JAX and PyTorch, targeting production-grade performance on GPUs/TPUs.

The role requires a PhD-level or senior researcher with a strong track record in numerical methods, AD, and scientific coding.

Qualifikationen

• PhD or equivalent research experience in SciML.
• Strong track record in differentiable numerical methods.
• Experience applying SciML to plasma physics is a plus.
• Development of production-grade solvers with testing and validation.

Aufgaben

• Design, implement and validate differentiable solvers in JAX for plasma physics problems (e.g., free-boundary equilibrium, transport, state estimation).
• Make solvers robust and fast for production use via advanced numerical techniques and GPU/TPU optimization.
• Use automatic differentiation for model calibration, parameter estimation and controller design with input from data/controls teams.
• Code ownership: testing, verification against references, packaging and documentation.

Kenntnisse

Differentiable solvers

SciML

ODE/PDE methods

GPU/TPU optimization

Automatic differentiation

Python

JAX

PyTorch

Julia

Ausbildung

PhD or equivalent research experience (SciML)

Tools

JAX

PyTorch

Julia

Diffrax/Equinox

Jobbeschreibung

Fusionality in Lausanne is seeking a Computational Scientist specialized in Scientific Machine Learning (SciML) to design differentiable solvers for fusion plasma problems. You will work close with control and physics teams to build end-to-end, differentiable solvers in JAX and PyTorch, targeting production-grade performance on GPUs/TPUs.

The role requires a PhD-level or senior researcher with a strong track record in numerical methods, AD, and scientific coding.