Search Safety Intern — LLM Training Data & Prompts

TikTok; San Jose (CA); Hybrid; USD 28,000 - 40,000; Full time

Benefits offered by this job

Health insurance

Life insurance

Wellbeing benefits

Paid holidays (10 per year)

Paid sick time

Housing allowance (where eligible)

Job summary

TikTok is seeking a Search Safety PM Intern in San Jose to support the development and training of next-generation large language models. You will build high-quality training resources, design prompts, and generate large-scale labeled datasets powering model learning and performance.

This role sits at the intersection of content understanding and AI, requiring strong language skills, attention to detail, and the ability to transform complex content into structured training inputs.

Qualifications

• Strong ability to understand, interpret, and structure complex textual content with high attention to detail.
• Exceptional English writing and communication skills, capable of producing clear, precise content at scale.
• Interest in large language models and ability to rapidly learn AI concepts and workflows.
• Self-driven and responsible, able to handle large volumes of work with quality.

Responsibilities

• Training corpus construction: build and maintain high-quality datasets for LLM training.
• Label system design & scaling: design labeling frameworks and generate large-scale labeled data.
• Data generation & curation: produce and refine training data per standards and use cases.
• Prompt writing & optimization: draft and refine prompts for training scenarios.
• Model behavior analysis: analyze outputs to identify gaps and improve prompts.
• LLM familiarity & application: apply understanding of LLM capabilities to data and prompts.

Skills

Content sensitivity

Analytical thinking

English proficiency

LLM awareness

Ownership & execution

Education

Undergraduate/Master's in CS/Statistics/Info Mgmt/Data Science

Job description

TikTok is seeking a Search Safety PM Intern in San Jose to support the development and training of next-generation large language models. You will build high-quality training resources, design prompts, and generate large-scale labeled datasets powering model learning and performance.

This role sits at the intersection of content understanding and AI, requiring strong language skills, attention to detail, and the ability to transform complex content into structured training inputs.