Research Scientist — Physics Foundation Models

Build scalable foundation models that learn from high-fidelity physics simulations and generalize across geometries, flow regimes, and operating conditions.

• Develop and train neural surrogate and foundation models for physical systems.
• Design pretraining, fine-tuning, and transfer-learning approaches for heterogeneous simulations.
• Explore neural operators, geometric and conditional model architectures.
• Study scaling with model size, data, fidelity, and compute.
• Implement distributed multi-GPU training workflows.
• Evaluate out-of-distribution geometries and physical conditions.
• Partner with simulation teams to define effective training datasets.

• Experience in deep learning research or large-scale training engineering.
• Strong Python and PyTorch or JAX.
• Distributed training, GPU optimization, and experiment tracking.
• Knowledge of numerical analysis, scientific computing, or physics-informed ML.
• Ability to design rigorous experiments and interpret results.

• Foundation models, neural operators, or scientific ML at scale.
• FSDP, DeepSpeed, mixed precision, and distributed data loading.
• Multi-fidelity modeling, domain adaptation, self-supervised learning.

Deliver reproducible scaling results and physics models with demonstrably useful generalization to unseen cases.

Send your CV and a relevant paper, technical project, GitHub repository, or portfolio to careers@ratiolabs.ai. Suggested email subject: “Physics Foundation Models”.