verified listingSign up to apply with your verified profile — no re-entering experience or references.
source · wttj·req · jb_b31911e7ec·listed 6d ago

Model Behavior Architect- Function Calling

Mistral Ai·London, England, United Kingdom·Hybrid·Full-time
Sourced listing · wttjNo salary disclosed
Posted
23 May 2026
via wttj
Type
Full-time
Arrangement
Hybrid
United Kingdom
Deadline
22 June 2026
closes in 25d
compensation · not disclosed
Salary not shared
Sign up to see our estimate based on role, location, and seniority.
source · estimate pending

Summary

the pitch

Join Mistral AI, a pioneering company in the AI industry. As a Model Behavior Architect on the Function Calling team, you will define and measure how large language models (LLMs) use tools, invoke functions, and orchestrate complex workflows. You will work closely with the Science team to establish evaluation criteria for function calling and improve model behavior. This role requires expertise in API design, structured outputs, and LLM agents. Mistral AI offers a competitive salary, equity, health insurance, transportation allowance, sport allowance, meal vouchers, generous parental leave policy, and visa sponsorship.

Role

posted by company

Join Mistral AI, a pioneering company in the AI industry. As a Model Behavior Architect on the Function Calling team, you will define and measure how large language models (LLMs) use tools, invoke functions, and orchestrate complex workflows. You will work closely with the Science team to establish evaluation criteria for function calling and improve model behavior. This role requires expertise in API design, structured outputs, and LLM agents. Mistral AI offers a competitive salary, equity, health insurance, transportation allowance, sport allowance, meal vouchers, generous parental leave policy, and visa sponsorship.

Key responsibilities

  • Interacting with models to identify areas for improvement in function calling and tool use behavior.
  • Designing and implementing evaluations, data guidelines, data generation, and synthetic tool environments and APIs.
  • Developing robust evaluation pipelines for the function-calling capabilities of model candidates.