Model Behaviour Architect- Multilingual

Mistral AI Mistral AI · AI Frontier · Paris, France · Research

Mistral AI is seeking a Model Behaviour Architect for their Multilingual team to define and measure model behavior across languages. This role involves working with AI Scientists to ensure language adherence, naturalness, cultural alignment, and consistent capability in multilingual models. Responsibilities include identifying areas for improvement, defining high-quality behavior per language, designing and implementing evaluations, data guidelines, and synthetic testing environments, and shaping how models reflect user cultural contexts. The ideal candidate has expertise in building robust evaluations, a strong understanding of linguistics or multilingual NLP, and experience with LLM agents, reasoning, and tool use.

What you'd actually do

  1. Define what high-quality behavior means per language
  2. Design and implement evals, data guidelines, data generation, and synthetic testing environments for multilingual settings
  3. Shape how models reflect the cultural context of the users and regions they serve
  4. Identify and fix edge-case behaviors through rigorous cross-language testing
  5. Develop robust evaluation pipelines for model candidates

Skills

Required

  • building robust evaluations
  • linguistics
  • language
  • translation
  • LLM agents
  • reasoning
  • tool use
  • multilingual NLP

Nice to have

  • Depth in linguistics or multilingual NLP is strongly preferred

What the JD emphasized

  • expert in model evaluation
  • policy writing
  • building eval pipelines
  • track record of delivering innovative, out-of-the-box solutions

Other signals

  • multilingual model behavior
  • model evaluation
  • language adherence
  • cultural alignment
  • cross-language testing