AIGT Agent Operations (Model Data Operations) - TikTok Platform Safety

About the Team

The Platform Responsibility team is focused on advancing content safety through practical AI innovation. The group is building next-generation infrastructure powered by advanced large language models to improve risk detection and governance. By combining strong engineering, high-quality data pipelines, and AI-driven decision-making, the team aims to improve efficiency, sharpen precision, and strengthen user trust across TikTok.

About the Role

This position sits within TikTok’s community-guidelines-driven safety work and centers on using large language models and prompt engineering to build an Agent capable of delivering highly accurate content-safety evaluation. The role will guide strategy operations and prompt refinement to improve the Agent’s ability to correctly identify risky content with strong recall.

Core Responsibilities

• Design, test, and debug prompts for specific risk scenarios, while keeping a versioned prompt library so the most suitable instruction set can be reused for each case.
• Build and manage a high-quality few-shot example library, ensuring coverage of newly emerging underground-industry patterns and difficult edge cases.
• Investigate false negatives and false positives in depth to determine whether the issue comes from unclear prompt instructions, stale examples, or policy ambiguity.
• Rapidly adapt interception strategies when public issues escalate or new attack patterns appear, then convert temporary responses into long-term operating capabilities after review.
• Maintain the external knowledge base used by the RAG system for political, current-affairs, and person-related risk areas so the Agent stays current with new terminology.
• Track daily performance metrics, prepare daily and weekly operation reports, and flag abnormal movements quickly so causes can be diagnosed and addressed.

Qualifications

• Bachelor’s degree or higher, along with at least 1 year of experience in content risk control, strategy operations, search suggestion strategy, or AI product operations.
• Strong understanding of content safety operations and the underlying business context.
• Practical knowledge of LLM principles and hands-on prompt engineering experience.
• Proven ability to improve model performance by tuning context, persona, and example selection.
• Strong data analysis skills and the ability to spot strategic problems quickly from metric changes.
• Self-driven, outcome-oriented, and able to work with stakeholders in fast-moving and ambiguous environments.

Additional Information

TikTok’s mission is to inspire creativity and bring joy through short-form mobile video. The company’s global headquarters are in Los Angeles and Singapore, with offices in New York City, London, Dublin, Paris, Berlin, Dubai, Jakarta, Seoul, and Tokyo.

Employees are encouraged to work with curiosity, humility, resilience, and an “Always Day 1” mindset, while embracing learning, innovation, and cross-team collaboration.

TikTok is committed to building an inclusive workplace that values different skills, experiences, and viewpoints, and reflects the global communities its platform serves.