Search Safety Project Intern (TikTok Safety Product) - 2026 Start (BS/MS)

Description du poste:

About the Team The Safety Product team is at the forefront of building and optimizing content safety systems. With a focus on optimising and advancing content safety, we leverage advanced large language models to enhance review efficiency, risk control, and user trust. Working closely with business and technical stakeholders, we deliver scalable solutions that keep pace with rapid global growth. We are seeking a passionate and detail-oriented Search Safety PM Intern to join our team and play a pivotal role in safeguarding the health and security of our search business. Your core mission will be to support the development and training of next-generation large language models. In this role, you will work closely with the team to build high-quality training resources, design effective prompts, and generate large-scale labeled datasets that directly power model learning and performance. This position sits at the intersection of content understanding, language, and applied AI. You will leverage strong analytical thinking and language skills to transform complex real-world content into structured, high-quality training data for LLMs. Your work will play a foundational role in shaping model behavior, alignment, and real-world usability. As a project intern, you will have the opportunity to engage in impactful short-term projects that provide you with a glimpse of professional real-world experience. You will gain practical skills through on-the-job learning in a fast-paced work environment and develop a deeper understanding of your career interests. Applications will be reviewed on a rolling basis - we encourage you to apply early. Successful candidates must be able to commit to at least 3 months long internship period. Responsibilities LLM Training Data & Knowledge Base Development - Training Corpus Construction: Assist in building and maintaining high-quality datasets and knowledge bases for new LLM training initiatives, ensuring accuracy, diversity, and contextual richness. - Label System Design & Scaling: Help design labeling frameworks and taxonomies, and generate large-scale, high-consistency labeled datasets to support supervised and reinforcement learning workflows. - Data Generation & Curation: Produce, review, and refine large volumes of training data based on defined standards, use cases, and evaluation criteria. Prompt Design & Model Interaction - Prompt Writing & Optimization: Draft, iterate, and optimize prompts for different training and evaluation scenarios, ensuring clarity, coverage, and alignment with model objectives. - Model Behavior Analysis: Analyze model outputs to identify gaps, biases, or failure patterns, and translate insights into improved prompts or data requirements. - LLM Familiarity & Application: Apply a strong understanding of LLM capabilities and limitations to design data and prompts that meaningfully improve model performance. Content Understanding & Quality Assurance - Content Interpretation: Deeply understand complex content across domains, identify key signals, intent, and nuances, and convert them into structured training inputs. - Quality Control: Conduct quality reviews on datasets, labels, and prompts, ensuring consistency, logical soundness, and adherence to guidelines. - Iteration & Improvement: Continuously refine data standards and workflows based on model feedback and project needs

Profil requis du candidat:

Minimum Qualifications - Actively enrolled university students who can commit to the role for at least 3 months and above ideally with a minimum of 3 working days per week. - Strong Content Sensitivity & Analytical Thinking: Excellent ability to understand, interpret, and structure complex textual content, with high attention to detail and nuance. - Outstanding English Proficiency: Exceptional English writing and communication skills, with the ability to produce clear, precise, and logically structured content at scale. - LLM Awareness & Learning Agility: Strong interest in large language models, with the ability to quickly learn AI-related concepts, tools, and workflows and apply them in practice. - Ownership & Execution Ability: Highly responsible, self-driven, and capable of handling large volumes of work with consistency and quality. Preferred Qualifications - Humanities / Social Sciences Background: Currently pursuing or recently completed a major in humanities, social sciences, foreign languages, international relations, or related fields. - Prior experience in content annotation, data labeling, research assistance, or AI-related operations. - Experience interacting with LLMs (e.g., prompt engineering, evaluation, or content generation projects). If you have any questions, please reach out to us at apac-earlycareers@tiktok.com