Mistral AI is looking for Applied Scientists with deep expertise in engineering sciences to work at the frontier of AI-accelerated simulation. You will work with industrial customers and internal research teams to build and deploy AI Physics Models alongside our existing offerings of Large Language Models (LLMs).

You will contribute across the full stack: curating high-fidelity simulation datasets, training and evaluating models, and delivering production-grade AI solutions directly to engineering teams. Target domains include computational fluid dynamics, structural mechanics, semiconductor design, multi-physics modelling, and digital twins.

Working cross-functionally with research, product, and customer-facing teams, you will ensure our models meet real engineering standards — not just benchmark metrics.

Design and run large-scale simulation campaigns using domain-specific solvers (e.g. OpenFOAM, ANSYS, COMSOL, Abaqus)
Run training of AI models on physics data, with rigorous evaluation of coverage, accuracy, and quality against industry validation standards
Build tools and frameworks for automated dataset creation, simulation pipeline management, and model evaluation
Develop agents and RAG that integrate LLMs with engineering simulation workflows
Collaborate closely with the collaborate with the science/research team on training runs and diagnose failure modes arising from data gaps or architecture limitations
Manage research projects and client communications with engineering teams

Qualifications

Fluent English with excellent communication skills - able to explain technical simulation concepts to both engineering and non-technical audiences
PhD or Master's in AI or an engineering science: Mechanical Engineering, Electrical Engineering, Computational Fluid Dynamics, Structural Mechanics, Semiconductor Engineering, or a related field. A solid understanding of deep learning and engineering or physics is a must.
Comfortable with PyTorch or JAX for implementing and training models
You write clean, readable Python code and are comfortable in Linux/HPC environments
Self-directed - you don't need detailed roadmaps to make progress
Low-ego, collaborative, and eager to learn at the intersection of simulation and ML
Demonstrated success through industrial projects, academic work, or personal projects
Have industrial or academic experience with simulation solvers (e.g. OpenFOAM, ANSYS, COMSOL, Abaqus, or equivalent)
Have applied ML methods to simulation or surrogate modelling
Have experience automating large-scale simulation campaigns on HPC clusters
Have contributed to a large open-source or industry codebase
Have publications in engineering or ML venues (NeurIPS, ICLR, etc.)