Nuclear Forensics
Enregistrez cette offre et organisez votre recherche
Créez un compte gratuit pour enregistrer des offres d'emploi, créer des alertes et revenir à cette liste depuis votre tableau de bord.
Mercor in Paris is assembling a panel of nuclear materials and safeguards experts to red-team frontier AI models. You will write prompts labeled across benign, dual-use, and adversarial levels, and evaluate responses against a defined policy standard.
This role emphasizes written justification and requires clear reasoning; prior technical writing or expert witness experience is valuable. You will also compose the reference answer with detailed rationale.
#J-18808-Ljbffr-
Nuclear Forensics Expert
Il y a 1 jour
Paris, Île-de-France Obsidian Temps pleinAbout the workMercor is assembling a panel of nuclear materials and safeguards experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.You
-
Nuclear Forensics Expert
Il y a 3 jours
Paris, Île-de-France Mercor Temps pleinAbout the workMercor is assembling a panel of nuclear materials and safeguards experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.You
-
Nuclear Safeguards Expert
Il y a 5 jours
Paris, Île-de-France Mercor Temps pleinAbout the work Mercor is assembling a panel of nuclear domain experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones. You will: Write
-
Nuclear Safeguards Expert
Il y a 5 jours
Paris, Île-de-France Obsidian Temps pleinAbout the work Mercor is assembling a panel of nuclear domain experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones. You will: Write
