Research Scientist Intern- (Trust and Safety - CV​/NLP​/Multimodal LLM PhD

Position: Research Scientist Intern- (Trust and Safety - CV/NLP/Multimodal LLM) - 2026 Start(PhD)
The Trust and Safety algorithm team is responsible for developing state-of-the-art computer vision, NLP and multimodality models and algorithms to protect our platform and users from the content and behaviors that violate community guidelines and related regulations. With the continuous efforts from our team, Tik Tok is able to provide the best user experience and bring joy to everyone in the world.

We are looking for talented individuals to join us for an internship in 2026. PhD Internships at Tik Tok aim to provide students with the opportunity to actively contribute to our products and research, and to the organization's future plans and emerging technologies.

In our team, you will have the opportunity to participate in the development of the cutting-edge content understanding model to help improve the recognition ability of violated content in Tik Tok, and will also be responsible for optimizing our distributed model training framework continuously.

PhD internships at Tik Tok provide students with the opportunity to actively contribute to our products and research, and to the organization's future plans and emerging technologies. Our dynamic internship experience blends hands-on learning, enriching community-building and development events, and collaboration with industry experts.

Applications will be reviewed on a rolling basis - we encourage you to apply early. Please state your availability clearly in your resume (Start date, End date).

Responsibilities
1. Develop computer vision model or multimodality model to recognize violation content in Tik Tok
2. Explore cutting-edge multimodal or computer vision large models (CLIP, COCA, ALBEF, BLIP, Flamingo, ViT-G, ViT-22B, EVA-enormous, etc)
3. Explore the application of LLM in our business scenarios, like pre-training, zero-shot/ few-shot learning, hard case mining, etc
4. Continuously optimize the training framework to better adapt to the training of large models.