Foundations & Intelligence Service, Trust & Safety is TikTok’s foundation-model team responsible for core development and research that push the boundaries of our LLM and VLM capabilities—with trustworthiness and safety built in from day one. Our work enables product teams and customers to adopt advanced AI smoothly and responsibly, with safety woven into the model stack rather than bolted on.
We are looking for talented individuals to join us for an internship. PhD internships at Our Company provide students with the opportunity to actively contribute to our products and research, as well as to the organization's future plans and emerging technologies. Our dynamic internship experience blends hands-on learning, enriching community-building and professional development events, and collaboration with industry experts. Applications will be reviewed on a rolling basis, so we encourage you to apply early. Please clearly state your availability in your resume (Start date, End date).
Responsibilities:
- Pretraining & Continued Pretraining (CPT): Explore and develop approaches that improve general capability and safety for LLMs and VLMs via pretraining/CPT.
- Evaluation & Measurement: Build advanced evaluation systems to study emerging LLM/VLM skills and safety behaviors, aligned with real downstream use.
- Post-Training & RL: Develop advanced reinforcement-learning strategies (e.g., RLHF/RLAIF/DPO variants), and investigate how to balance pretraining and post-training for better capability and safety alignment.
- Scaffolded Settings & Agents: Probe new failure modes and pitfalls in scaffolded environments (agents, workflows, tool use), and translate insights into robust mitigations.
Minimum Qualifications:
- Currently pursuing an PhD in Computer Science or a related technical field.
- Research experience in at least one of: LLMs, AI Safety, Computer Vision, Multimodality.
- Actively track recent AI-safety developments (familiar with current papers, benchmarks, and terminology).
- Proficient with at least one deep learning framework (e.g., PyTorch, TensorFlow).
- Excellent analytical/problem-solving ability, clear logical thinking, and strong communication/collaboration skills.
Preferred Qualifications:
- Publications in top AI or Security venues (e.g., NeurIPS, ICLR, ICML, COLM, CVPR, ICCV, ECCV, USENIX Security, ACM CCS, IEEE S&P, NDSS).
By submitting an application for this role, you accept and agree to our global applicant privacy policy, which may be accessed here: https://careers.tiktok.com/legal/privacy