Database Administrator - Graduates - AI Training - Dunedin, Zealand

Contribute database and records expertise to the training and evaluation of AI systems through paid, project-based tasks. Work may involve reviewing generated records, testing how models handle difficult inputs, and labeling examples against data quality standards. Applicants complete a short skills assessment before they can be invited to participate.

• Check AI-generated entries and extracted data for accuracy, completeness, and consistent formatting.
• Identify classification, labeling, and field-mapping errors in generated datasets, and record issues against quality guidelines.
• Create realistic examples of messy, ambiguous, or conflicting records to test model behavior.
• Label and classify samples to reinforce data structures and validation rules.
• Compare outputs with professional data-entry standards and flag inconsistencies.

• Professional experience in high-volume data entry, data processing, database management, or records administration.
• Strong accuracy and attention to detail when working with large datasets, including spotting duplicates and formatting problems.
• Ability to process structured and unstructured information efficiently while maintaining quality.
• Familiarity with data formats and validation rules, and ability to recognize logical or factual errors in AI outputs.
• Written English sufficient to assess clarity and correctness.
• A Pay Pal account to receive task payments.

• Experience with spreadsheets or database tools such as Excel, Google Sheets, Airtable, SQL, or Access; CRM, ERP, or document management systems; and documentation platforms such as Confluence or Notion.
• Additional languages, particularly for reviewing localized data.

Tasks are paid, with rates varying by project; the listing says some researchers may pay up to $25 per hour. Some tasks require an uninterrupted hour, though shorter tasks may also be available. Work is described as flexible and remote, subject to available assignments.