内容质量与评估专家-TikTok
Core
Build and iterate the core content safety evaluation dataset (Golden Set) to provide high-accuracy ground truth for platform safety strategies, AI model assessment, and global execution teams.
Role type
Senior Content Safety Quality Expert (AI Data Evaluation & Strategy)
Builds
Content safety evaluation datasets, standardized operational guidelines (SOPs), and quality control mechanisms
Domain
Internet content safety, AI model evaluation, global operations
Deliverable
production ML models
Required skills
Content safety domain expertise, AI data evaluation, logical reasoning, abstraction of complex rules, cross-functional collaboration, data metric tracking, root cause analysis
Preferred skills
Machine learning logic, data annotation frameworks, second language proficiency (Japanese, Thai, Korean, German)
Responsibilities
Construct and iterate the Golden Set by mining long-tail and high-risk edge cases; Analyze high-risk and controversial safety cases to identify misjudgment patterns and risk trends; Lead the abstraction of complex rules into executable SOPs; Drive the content safety data closed-loop by collaborating with operations, algorithms, and product teams to identify vulnerabilities and improve human quality and Agent interception efficiency; Establish and refine daily quality control mechanisms and track core data metrics to improve accuracy and operational efficiency.