Research Engineer - AI Safety Benchmarks

Visa Hunt; Paris; Hybride; EUR 85 000 - 125 000; Plein temps

Avantages offerts par ce poste

Paid time off

Hybrid Paris work with relocation

France medical insurance

All hardware and tools

Subscriptions for AI tools

Team off-sites

Résumé du poste

White Circle is seeking a Research Engineer to build and maintain internal benchmarks across single and multi-turn content, focusing on agentic safety. You will collaborate with the team on studies of agent behaviors in real-world settings and help shape evaluation methodologies for our safety platform.

You will write Python that others can extend, optimize inference setups, and contribute to production-grade experimentation.

Qualifications

• Candidates should write clean, maintainable Python code.
• Experience building and maintaining benchmarks for LLMs.
• Strong understanding of AI safety and evaluation methodologies.
• Ability to productionize research code in a scalable way.

Responsabilités

• Own and maintain the internal benchmark suite for single/multi-turn content guardrails and agentic safety.
• Build benchmarks that distinguish model capabilities.
• Collaborate with product to build evals for flagship models.
• Extend evals to new features and data domains.
• Study and quantify realistic agentic and LLM failure modes in the wild.

Connaissances

Python

LLM benchmarking

AI safety

Production code

Inference orchestration

Frontier models

Red-teaming

Agentic benchmarks

Description du poste

White Circle is seeking a Research Engineer to build and maintain internal benchmarks across single and multi-turn content, focusing on agentic safety. You will collaborate with the team on studies of agent behaviors in real-world settings and help shape evaluation methodologies for our safety platform.

You will write Python that others can extend, optimize inference setups, and contribute to production-grade experimentation.