Bilingual AI Trainer EN/DE - LLM Evaluation & RLHF

Rex.zone; United States; À distance; USD 48 000 - 55 000; Plein temps

Résumé du poste

Rex.zone is seeking Germany-based bilingual (English/German) AI Generalist Trainers to support LLM evaluation, RLHF ranking, and QA evaluation workflows. You will assess outputs, write rationales, and ensure compliance with annotation guidelines in both English and German.

The role is full-time, remote, with compensation of USD 35–40 per hour, based on performance and project needs. You will maintain throughput and inter-annotator consistency while improving data quality.

Qualifications

• Must be based in Germany and eligible to work remotely per local requirements.
• Full professional fluency in English and German (reading and writing).
• Strong analytical reasoning and attention to detail.
• Ability to follow rubric-based guidelines and explain decisions in writing.

Responsabilités

• Perform large language model evaluation of model-generated outputs in English and German
• Rank multiple responses using defined rubrics and reasoning standards
• Write concise rationales that justify selections and support RLHF learning signals
• Execute QA evaluation for correctness, consistency, and annotation guidelines compliance
• Validate ambiguous cases, identify policy gaps, and elevate issues with evidence
• Apply prompt evaluation for instruction-following, helpfulness, and factuality checks
• Support content safety labeling where required and document decisions for auditability
• Maintain training data quality by meeting throughput targets and inter-annotator consistency goals

Connaissances

Bilingual English/German

Analytical reasoning

Attention to detail

Description du poste

Rex.zone is seeking Germany-based bilingual (English/German) AI Generalist Trainers to support LLM evaluation, RLHF ranking, and QA evaluation workflows. You will assess outputs, write rationales, and ensure compliance with annotation guidelines in both English and German.

The role is full-time, remote, with compensation of USD 35–40 per hour, based on performance and project needs. You will maintain throughput and inter-annotator consistency while improving data quality.