AI Agent Operations Specialist （Model Data Operations) - TikTok Platform Safety

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

Prompt engineering and operations for an LLM-based agent that evaluates TikTok content for safety risks. The work covers writing and tuning prompts, maintaining few-shot samples, and analyzing missed and wrongly removed content. It sits in the Platform Responsibility team, which is building AI-driven risk detection and governance.

What you'd do

• Write and debug prompts for specific risk scenarios and maintain the prompt version library.
• Filter and maintain few-shot examples, including recent underground industry variants and edge cases.
• Analyze false negatives and false positives to find whether prompts, samples, or policy caused the error.
• Quickly adjust interception prompts when public opinion events or new attack types emerge, then make lasting changes in post-event review.
• Maintain the external knowledge base for the RAG system covering political, current affairs, and person-related risks.
• Monitor metrics daily, produce daily and weekly reports, and investigate abnormal fluctuations.

What it asks for

• A bachelor's or higher, plus at least a year in risk control, strategy ops, search suggestion work, or AI product ops.
• Deep understanding of the content safety business.
• Hands-on prompt work with LLMs, with a record of lifting model results by tuning context, persona, and examples.
• Strong data analysis skills for spotting strategy issues from metric changes.
• Skill at winning over stakeholders when things are moving fast and unclear (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.