Research Scientist (AI Behaviours)

Responsibilities

Our client, a confidential AI safety organization based in Paris, is seeking a Research Scientist to investigate how large language model agents fail in real-world conditions. You will design and execute empirical studies to identify deception, misalignment, and unsafe behaviors in concrete experimental settings.

In this role, you will:

• Lead research projects from initial concept through execution—transforming vague safety concerns into falsifiable experiments with clean baselines and defensible results
• Develop automated audit systems that discover and characterize problematic model behaviors at scale
• Study how misalignment and bias manifest when real users interact with AI agents, converting findings into production-ready evaluations
• Pressure-test frontier AI agents in realistic, high-stakes scenarios to identify failure modes before deployment
• Conduct both white-box and black-box investigations to understand model failure mechanisms
• Publish findings as blog posts and conference papers, while feeding insights back into internal safety systems

What We're Looking For

Required qualifications:

• Demonstrated track record of empirical research in agent behavior, model evaluation, alignment, or closely related areas
• Strong ML engineering capabilities—you can independently build research prototypes involving fine-tuning, agent inference, and evaluations
• Proven experimental design skills: isolating agent failure modes, calibrating judges and baselines, and distinguishing signal from artifact
• Ability to transform vague behavioral questions into well-defined experiments and iterate quickly
• Fluency with frontier models and AI coding agents as a power-user

Preferred qualifications:

• Published research at top-tier venues (NeurIPS, ICML, ICLR, ACL, or similar)
• Experience with interpretability tooling and concepts (neural language analysis, sparse autoencoders, persona vectors, etc.) and ability to conduct white-box model investigations
• MSc or PhD in machine learning, computer science, cognitive science, computational neuroscience, physics, or related quantitative field
• Participation in AI safety fellowships (MATS, ASTRA, Anthropic Fellows, etc.) or comparable self-directed research experience

How We Work With You

As your recruiting partner, Get A Job.ai represents you throughout the entire hiring process. Here's how it works:

• Apply directly through the Get A Job.ai platform
• Our recruiting team will screen your application and conduct an initial assessment
• We submit qualified candidates to our client and coordinate all interview scheduling
• Please do not contact the employer directly—all communication flows through our team to ensure the best candidate experience

The client's interview process typically includes an introductory call, a take-home research task, a technical interview with the research leadership, and a final conversation with executive leadership.

Location: Paris, France (hybrid work arrangement). Relocation package may be available for qualified candidates.

Pay

Compensation details will be discussed during the screening process with our recruitment team.

Equal Employment Opportunity: Get A Job.ai is committed to fair and equitable recruiting practices. We evaluate all candidates based on qualifications and merit, and we encourage applications from individuals of all backgrounds.