Germany-Based English & German AI Generalist Trainer (Remote, Full-Time)

Rex.zone is hiring Germany-based English/German AI Generalist Trainers to support RLHF and large language model evaluation by assessing, ranking, and quality-checking model outputs to strengthen training data quality and drive model performance improvement.

About The Role

In this role, you will evaluate and improve AI/LLM behaviors through RLHF-style workflows. You will review prompts and model responses, validate factuality and instruction-following, apply annotation guidelines compliance standards, and write clear rationales that improve training data quality.

Key Responsibilities

• Perform large language model evaluation across English and German prompts and outputs
• Rank multiple model-generated responses using defined rubrics and consistent reasoning
• Conduct QA evaluation checks by auditing items for annotation guidelines compliance
• Write concise rationales that justify rankings and identify reasoning errors
• Validate outputs for safety, policy adherence, and content safety labeling needs
• Identify edge cases and failure modes; document findings for model performance improvement
• Apply data labeling standards to create reliable feedback signals for RLHF pipelines
• Track discrepancies, elevate guideline questions, and support consistent quality targets

Basic Qualifications

• Must be based in Germany and able to work remotely from Germany
• Fluency in English and German (strong reading and writing skills in both)
• Strong analytical skills; ability to evaluate arguments, logic, and evidence
• Exceptional attention to detail and consistency in applying rubrics
• Comfortable working with sensitive or policy-relevant content under safety labeling rules

Preferred Qualifications

• Experience with AI data labeling, RLHF, prompt evaluation, or LLM evaluation
• Familiarity with LLM failure patterns (hallucinations, instruction drift, bias, safety issues)
• Background in linguistics, translation, writing/editing, QA, research, or content moderation

Compensation

Hourly pay range: $30–$50 USD per hour (base), based on skills, language proficiency, and evaluation quality.