French AI Data Trainer - Remote

The position of French AI Data Trainer is a fully remote contract role focused on enhancing the safety and reliability of artificial‑intelligence systems. The company works on large language models and needs a specialist who can assess AI‑generated replies and add safety‑oriented annotations in both French and English. Day‑to‑day duties involve reviewing machine‑produced text for correctness, adherence to safety rules, and policy alignment. The contractor will label datasets with precise French and English language, flag content that includes hate, sexual material, violent or self‑harm themes, biased or misleading information, illegal actions, and malicious code. They will also compare multiple model outputs, assign scores using quality and safety rubrics, detect patterns of model failure such as evasion or escalation, and conduct adversarial tests to expose vulnerabilities. Clear documentation of findings and contribution to the improvement of annotation guidelines are required to keep policy interpretation uniform. Essential qualifications include a bachelor’s degree or comparable experience in fields such as linguistics, communications, psychology, law, policy, or security studies, native‑level French reading and writing, and at least C1‑level English proficiency. The applicant must have a background in trust and safety, content moderation, policy enforcement, or risk operations, as well as proven experience in testing large language models and reporting results. Strong analytical judgment, meticulous attention to detail, and the ability to articulate decisions clearly are mandatory, along with comfort working independently in a remote contractor setting. Desirable experience covers prior work with AI training data, annotation or model assessment, familiarity with platforms like ChatGPT, Gemini or Perplexity, and knowledge of structured safety taxonomies. A history of investigative or policy‑driven analytical roles is also a plus.