Research Engineer (AI Safety)

Location: Hybrid work from the Paris or London offices. Relocation support is available for moves to Paris after the probationary period.

Company

hirify.global is an AI Safety company building a safety, reliability, and optimization layer for AI systems through natural-language policies, automated testing, enforcement, and continuous improvement.

What you will do

• Own and maintain internal benchmarks for single- and multi-turn content guardrails and agentic safety.
• Build benchmarks that distinguish specific model capabilities with measurable and defensible results.
• Work with the product team to create evaluations for flagship model functionality.
• Develop benchmarks for new research features and adapt evaluations to new verticals and changing product data.
• Study and quantify realistic agentic and LLM failure modes in real-world settings.

Requirements

• Experience building an LLM benchmark from scratch that distinguishes specific model capabilities.
• Experience creating synthetic data for post-training textual or multimodal models.
• Ability to reproduce published benchmark results and identify fragile or misleading methodology.
• Production Python experience, including writing maintainable code and designing extensible abstractions.
• Ability to build efficient LLM inference setups with parallel-call orchestration, retries, and rate-limit handling.
• Strong practical familiarity with frontier models and coding agents.

Nice to have

• Automated red-teaming experience.
• Experience with agentic scaffolds and public benchmark reproduction.
• Knowledge of reward-model, monitoring, and safety benchmarks.
• Published papers in evaluations or safety evaluation.

Culture & Benefits

• Small, focused team working on AI safety research and production systems.
• Competitive compensation including equity.
• Flexible time off and flexible hybrid work.
• Premium private health insurance and mental health support, including therapy coverage.
• Office meals, learning and development support, and required hardware and software.
• Team off-sites twice a year.

Hiring process

• Introductory call with the Talent Team.
• Test assignment.
• Technical interview with the Head of Fundamental Research, followed by a final interview with the CEO.