AI Safety Evaluation & Governance Product Manager Intern (TikTok-Platform Responsibility-Feed Safety) - 2027 Summer

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

This internship covers evaluation and governance of AI models that make content safety decisions for TikTok's feed. The team defines safety standards for AI systems, works with algorithm teams on model accuracy across safety scenarios, and builds the evaluation frameworks and ground truth datasets that show whether those models are reliable at scale.

What you'd do

• Design and run evaluation plans for AI safety models, defining objectives, metrics, and what good performance means per use case.
• Build and maintain ground truth datasets, including sampling strategies, data cleaning pipelines, and labeling consistency checks.
• Analyze model performance with statistical methods such as sampling design, confidence intervals, and error analysis.
• Turn evaluation findings into model improvement directions with algorithm engineers, including prompt design and model configuration iteration.
• Brief Policy, Operations, and Algorithm partners on evaluation results and governance standards, and help set shared quality expectations.
• Improve evaluation processes by finding gaps and proposing methodology upgrades as models and policies change.

What it asks for

• Solid applied statistics: sampling, hypothesis testing, confidence intervals, and distribution analysis.
• Foundational understanding of AI/ML concepts such as classification, NLP, LLMs, and precision/recall.
• Interest in governance, policy, or content safety.
• Working fluency in both English and Chinese.

Worth knowing

• Internship position.

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.