Italian AI Data Auditor & Transcription QA Specialist

Mercor; Roma; Sur place; EUR 19 000 - 25 000; Plein temps

Résumé du poste

Obsidian is seeking native or near-native Italian speakers to audit AI training data in their language. You will assess transcripts for accuracy and apply a fixed set of error codes, giving a concise rationale for each decision.

You will review audio with word-level segmentation, verify timestamps, adjust boundaries, and correct transcripts where needed. Calibration precedes production work, ensuring consistency and precision.

Qualifications

• Native Italian as spoken in Italy is required.
• Strong English reading and writing skills are required.
• Disciplined rubric application across judgments is essential.
• Careful listening to distinguish subtle sounds and errors.
• Prior AI data annotation or model evaluation experience is a plus.

Responsabilités

• Audit AI training data in Italian across two audit collections.
• Apply error codes, write a rationale for each judgment, and ensure consistency.
• Verify word-level timestamps and reference transcriptions in forced alignment tasks.
• Provide clear feedback following the rulebook for every task.

Connaissances

Italian fluency

English proficiency

Rubric discipline

Careful listening

Audio data annotation (plus)

Outils

Audio editing tools

Description du poste

Obsidian is seeking native or near-native Italian speakers to audit AI training data in their language. You will assess transcripts for accuracy and apply a fixed set of error codes, giving a concise rationale for each decision.

You will review audio with word-level segmentation, verify timestamps, adjust boundaries, and correct transcripts where needed. Calibration precedes production work, ensuring consistency and precision.