AI Scientist - Domain Expert Crashworthiness

Mistral

7d•Remote

About The Position

Mistral AI is looking for a Domain Expert in crashworthiness simulations to help build AI-accelerated simulation capabilities for industrial engineering. You will work at the intersection of industrial simulation, physics modeling, machine learning, and engineering workflows. Your role is to bring deep solid-mechanics expertise into the design, training, evaluation, and deployment of AI Physics Models for real engineering use cases. This is a hands-on technical role. We are looking for someone who understands industrial structural simulation not only conceptually, but through direct experience with simulation models, solver workflows, data generation, validation, and engineering decision-making. You will work with research, product, and customer-facing teams to ensure that our models are useful against real engineering standards — not only benchmark metrics. Relevant application areas may include automotive crashworthiness, aerospace structures, consumer-electronics drop/reliability, manufacturing, durability/fatigue, and other nonlinear structural mechanics problems.

Requirements

Deep expertise in crashworthiness, solid mechanics, and structural mechanics, with substantial experience in industrial simulation workflows.
Master’s degree or equivalent technical depth in mechanical engineering, aerospace engineering, civil/structural engineering, computational mechanics, applied physics, or a related field.
4+ years of relevant industrial experience (or PhD +1 years) in domains such as automotive, aerospace, and consumer electronics.
Hands-on experience with explicit dynamics for crash or impact simulation and understand nonlinear FEM and related topics such as contact, plasticity, structural dynamics, buckling, material modeling, fracture/damage, fatigue, crashworthiness, or durability.
Direct experience with simulation validation, correlation, model quality, numerical sensitivity, and engineering KPI definition.
Strong Python developer who can build and maintain reliable tools for simulation automation, data processing, and model evaluation.
Sound software engineering practices, including version control with Git, automated testing, code review, and clear documentation.
Hands-on experience working in Linux and HPC environments, including submitting and monitoring batch jobs, selecting appropriate compute resources, and troubleshooting simulation workflows.
Ability to run simulation campaigns efficiently across a compute cluster.
Comfortable operating in ambiguous technical environments and turning poorly defined industrial problems into scoped datasets, experiments, metrics, and execution plans.
Clear communication with both deep technical experts and non-specialist stakeholders.

Nice To Haves

Experience applying machine learning, surrogate modeling, reduced-order modeling, optimization, or data-driven methods to simulation problems.
Contributed to reusable internal tools, open-source code, simulation automation frameworks, or production-quality engineering workflows.
Experience working directly with industrial customers, product teams, or engineering decision-makers.
Publications, patents, internal technical leadership, or recognized contributions in engineering, simulation, computational mechanics, or ML-for-physics communities.

Responsibilities

Work with our research team to define, generate, and iteratively improve simulation datasets for training and evaluating crashworthiness foundation models, balancing coverage of relevant physical scenarios, simulation fidelity, and computational cost.
Design and run high-fidelity simulation campaigns using structural mechanics solvers such as Abaqus, LS-DYNA, Ansys Mechanical, Radioss or equivalent tools.
Define the relevant variation space for training data: geometry, mesh resolution, material behavior, boundary conditions, loading, contact, joints, failure modes, and engineering KPIs.
Build or guide automated pipelines for simulation setup, execution, post-processing, dataset creation, and model evaluation.
Work with research teams to train and evaluate AI models on simulation data, and diagnose failure modes caused by data gaps, poor coverage, numerical artifacts, or model limitations.
Evaluate model outputs against industrial engineering needs, including field-level accuracy, scalar KPIs, deformation modes, load paths, energy absorption, stress/strain fields, failure indicators, uncertainty, and out-of-domain behavior.
Work with industrial customers to understand their simulation workflows and engineering priorities, define use cases and success criteria, and incorporate their feedback into model development and validation.