Content Quality and Evaluation Specialist (Japanese Speaking) - Safety Model Operations

Written by the T&S Network from TikTok's posting.

Quality assurance and evaluation work on TikTok's automated content safety models, sitting in the Safety Model Operations team that builds and maintains the ML models and processes behind Trust & Safety systems. The role builds the ground-truth Golden Sets used for platform safety policy, AI model evaluation and enforcement teams, and covers the Japan market.

What you'd do

• Build and iterate Golden Sets of long-tail, high-risk and complex edge cases to set labeling ground truth.
• Run deep-dive reviews and root cause analyses on high-risk and heavily debated safety cases.
• Find misapplication patterns and risk trends, and turn macro policies into executable SOPs and guidelines.
• Work with Operations, Algorithm and Product teams to close the loop between Golden Set metrics and review and AI interception quality.
• Build daily QA mechanisms and quality management methods, and track data accuracy and efficiency metrics.

What it asks for

• Proven experience in internet Trust & Safety, QA, AI data evaluation or policy operations.
• Working fluency in English sufficient for professional use.
• Japanese proficiency, required for daily liaison with the local Japan team.
• Strong logical reasoning and synthesis skills for ambiguous, complex cases.
• Extreme attention to detail and consistency in standards.
• Understanding of how machine learning models work and how labeling schemes are built (preferred).

Our best summary of the posting, and it can be wrong. The employer's posting is what counts: read it in full on TikToknorth_east before you apply. If we got something wrong, report this listing.