AI Scientist, Agentic CAD (AI4Engineering)
Build training data, verifiers, benchmarks, and agent architectures that help Mistralβs language models reliably plan and execute complex engineering workflows across CAD, CAE, and semiconductor/EDA.
Weβre looking for researchers and engineers who love training the best AI models. Youβll work on the full training stack, from data and recipes to post-training, RL, and evaluation, alongside a small, high-caliber team building at the frontier of AI systems.
Pre-training, mid-training, and post-training of large models
Reinforcement learning, reasoning, and reward modeling
Data, evaluation, and the research behind making models measurably better
Have a track record of research or engineering on large-model training
Care about shipping results, not just publishing them
Are curious, direct, and comfortable with ambiguity
Build training data, verifiers, benchmarks, and agent architectures that help Mistralβs language models reliably plan and execute complex engineering workflows across CAD, CAE, and semiconductor/EDA.
Develop and deploy deep learning models for live surgical video and preoperative medical images, turning clinical data into real-time perceptual intelligence in collaboration with software, algorithm, and clinical teams.
Design YAML-based adversarial prompt injection tasks for an AI security evaluation platform, test them against frontier models, and validate scenarios that reveal whether agents can be manipulated into violating safety policies.
Develop AI agents and models that uncover and validate financial opportunities in complex healthcare data, turning manual analyses into scalable workflows with measurable value for health systems.
Own and evolve the AI model at the center of an Earth observation system, from defining learning objectives and running experiments to scaling training and integrating models into customer-facing products.