AI Quality Control Specialist
Pubblicato il 21-08-2026 - Educazione, formazione e pedagogia in Italia
Are you ready for new challenges and new opportunities?
Join our team
AI Quality Control Specialist – Korean Language
About the Role:
TrustScale is building scalable, high-quality human-in-the-loop systems to power AI. We are looking for an AI Quality Control Specialist - Korean Language who can analyze, validate, and challenge AI-generated outputs with precision and consistency. This is not a proofreading role. You will act as a quality gatekeeper, ensuring outputs meet standards for accuracy, utility, and linguistic quality across large-scale workflows.
What You Will Do:
Evaluate AI-generated content in Korean across multiple task types (annotation, review, validation).
Apply structured quality frameworks to assess:
Accuracy and factual correctness.
Utility (alignment with user intent).
Participate Clients Calls
Identify and flag:
Hallucinations and misinformation.
Cultural or linguistic mismatches.
Provide clear, structured feedback to improve upstream quality.
Detect patterns of errors across batches and contribute to quality insights.
Support calibration efforts to ensure consistent scoring across teams.
Contribute to guideline refinement and evaluation standards.
Who You Are:
Professional proficiency in Korean and strong command of English are required.
Highly analytical with strong critical thinking skills.
Comfortable working with ambiguous and evolving guidelines.
Detail-oriented with the ability to maintain consistency at scale.
Clear communicator, able to justify decisions and provide actionable feedback.
Naturally skeptical—able to question outputs and validate information.
Preferred Background:
Journalism, editorial, or investigative research.
Linguistics, translation, or localization QA.
Content moderation / Trust & Safety.
Experience in AI data annotation, evaluation, or QA.
What Success Looks Like:
High consistency in quality scoring across tasks.
Strong alignment with QA benchmarks.
Ability to detect non-obvious quality issues.
Feedback that improve