Frontier AI Safety Evaluator

Obsidian; Warszawa; On-site; PLN 180,000 - 300,000; Full time

Job summary

Obsidian in Warsaw seeks experienced AI Safety Practitioners to evaluate frontier models for safety, quality, and alignment across policy-sensitive topics. You will assess AI-generated responses, apply safety policies, and provide structured feedback to help improve model behavior through formal evaluations and collaboration with AI researchers and safety teams.

Responsibilities include identifying unsafe outputs, bias, and policy violations, and refining evaluation rubrics for RLHF, SFT, and

Qualifications

• Bachelor's degree or higher in a related field.
• 5+ years of professional AI Safety or related experience.
• Excellent written English, critical thinking, and analytical reasoning.

Responsibilities

• Evaluate AI-generated responses for safety, factual accuracy, policy compliance, and overall quality.
• Review content involving misinformation, political persuasion, self-harm, violence, cyber, biosecurity, and other sensitive domains.
• Apply and refine evaluation rubrics for RLHF, SFT, and AI safety benchmarking.
• Identify unsafe outputs, hallucinations, reasoning failures, and policy violations.
• Provide structured feedback to improve model alignment and safety performance.
• Collaborate with AI researchers and safety teams on ongoing evaluation initiatives.

Skills

Written English

Critical thinking

Analytical reasoning

Policy evaluation

Education

Bachelor's degree or higher

Tools

RLHF

SFT

AI evaluation

Job description

Obsidian in Warsaw seeks experienced AI Safety Practitioners to evaluate frontier models for safety, quality, and alignment across policy-sensitive topics. You will assess AI-generated responses, apply safety policies, and provide structured feedback to help improve model behavior through formal evaluations and collaboration with AI researchers and safety teams.

Responsibilities include identifying unsafe outputs, bias, and policy violations, and refining evaluation rubrics for RLHF, SFT, and