AI Safety Research Scientist — Adversarial Environments

White Circle; Paris; Hybride; EUR 60 000 - 90 000; Plein temps

Avantages offerts par ce poste

Paid time off

Comprehensive medical insurance

Covered subscriptions for AI tools

Team off-sites twice a year

Résumé du poste

White Circle in Paris is seeking a research scientist to build large-scale environments for testing LLM agents. You will conduct experiments to observe and catalogue agent failures, contributing to the development of AI safety protocols.

The ideal candidate has experience in empirical research and software engineering, preferably holding a MSc or PhD in a related field. The position is hybrid, with a relocation package available for qualified candidates.

Qualifications

• Experience in building non-trivial agent environments or automated research pipelines.
• Strong software and AI engineering capabilities.
• Experience with empirical research methodologies.
• Ability to define research questions and turn them into experiments.
• Fluent with AI models and coding.

Responsabilités

• Build adversarial environments for agents.
• Catalogue concrete agent failure modes.
• Run experiments with external APIs and internal models.
• Build realistic multi-agent environments.
• Turn findings into public writeups.

Connaissances

Agent environment development

Software engineering

Empirical research

AI model usage

Research pipeline orchestration

Formation

MSc or PhD in a related field

Outils

Python

AI frameworks

Description du poste

White Circle in Paris is seeking a research scientist to build large-scale environments for testing LLM agents. You will conduct experiments to observe and catalogue agent failures, contributing to the development of AI safety protocols.

The ideal candidate has experience in empirical research and software engineering, preferably holding a MSc or PhD in a related field. The position is hybrid, with a relocation package available for qualified candidates.