Job description
White Circle • Paris or London
Paris, France, London, UK
$150,000 - $250,000
- In this role, you'll study how LLM agents fail and misbehave in realistic scenarios, designing experiments to understand misalignment, deception, and unsafe behaviour.
- Develop automated audit agents that discover and characterise suspect model behaviour at scale.
- Analyse how misalignment and bias emerge when real users interact with agents and build evals for products.
- Pressure-test frontier models in high-stakes scenarios to identify failure modes before deployment.
- Conduct white-box and black-box investigations and publish findings as research papers and blog posts.
White Circle helps companies test, protect, observe, and improve AI interactions.