Research Scientist Graduate (Business Integrity) - 2027 Start (PhD)

Team Introduction We are the Business Integrity team. We put users at the center of everything we do and are committed to solving complex challenges in global business with cutting-edge AI technologies. As TikTok continues to grow rapidly worldwide, the team is building a next-generation content understanding and generation system for commercialization. We are now hiring Large Model/Agentic AI Research Scientists globally. We look forward to working with you to explore the algorithmic frontiers of large language models and multimodal technologies, publish influential research at top conferences, and translate groundbreaking research into a core engine for the next generation of commercial ecosystems.

We are looking for talented individuals to join our team. As a graduate, you will get opportunities to pursue bold ideas, tackle complex challenges, and unlock limitless growth.

Successful candidates must be able to commit to an onboarding date by the end of the year. Please state your availability and graduation date clearly in your resume.

- Research on foundation models and frontier generative technologies: Explore the capability boundaries of large language models (LLMs) and multimodal large models (MLLMs), especially the evolution of foundation capabilities for agentic scenarios. Conduct in-depth research and propose innovative post-training paradigms, tackling challenges in complex reasoning such as CoT reasoning and long-horizon planning, tool use, reinforcement learning alignment such as innovative applications of PPO, GRPO, and other new RLHF algorithms for agent behavior alignment, reward model design, and AI safety. Build model capabilities for autonomous decision-making and reflection from the underlying architecture. Explore and break through the underlying architectures of frontier visual generation algorithms such as diffusion and flow matching, building a strong perception and environment simulation foundation for multimodal agents. - Frontier breakthroughs in large model agents: Lead underlying algorithmic innovation for Agentic AI. Focus on academic challenges such as agentic reinforcement learning, complex logical reasoning and planning, long-term memory mechanisms, and knowledge injection. Explore the theoretical foundations of emergent multi-agent collaboration and optimal interaction strategies in highly complex, open-ended environments. - Forward-looking technical evaluation and theoretical construction: Define and build the evaluation system for next-generation AIGC and agent systems. Identify fundamental problems with core academic value in real-world, large-scale commercial scenarios such as advertising and e-commerce content ecosystems, and design new algorithmic frameworks to address long-tail generalization, interpretability, and multimodal alignment challenges. - Building academic influence: Continuously track and lead the technical frontier in AI, and strengthen the team’s technical influence.