Research Scientist, AI Safety - Global Frontier Tech Recruitment Program - 2027 Start (PhD)

We are looking for talented individuals to join our team in 2027. As a graduate, you will get opportunities to pursue bold ideas, tackle complex challenges, and unlock limitless growth. Launch your career where inspiration is infinite at our Company.

Successful candidates must be able to commit to an onboarding date by end of year 2027. Please state your availability and graduation date clearly in your resume.

Team Introduction: Our team provides end-to-end security solutions for ByteDance's large language model (LLM) business lines. We identify and mitigate emerging risks across the entire lifecycle of model training, inference, and deployment, while also developing tools and systems to safeguard LLM assets and related applications.

By advancing the deployment of AI security technologies in real-world business scenarios, we have built an industry-leading defense framework to support the safe evolution of artificial intelligence technologies.

Topic Content: While agent-based applications are experiencing explosive growth, they have also introduced new security challenges. The expanded attack surface covering data, decision-making, execution, and supply chains, combined with their intricate interactions and complex infrastructure, has rendered traditional security technologies ineffective.

This topic aims to systematically research adversarial testing and evaluation technologies for foundation models and agents in the Agentic AI era, as well as a full-link defense system based on foundation model trusted privacy computing, to ensure the safe and stable development of the company's business related to foundation models, including:

1. Attack-defense detection and evaluation methods, testing benchmarks, and toolchains for multimodal foundation models and agents;
2. Risk defense technologies and trustworthy runtime security assurance for agent-based applications;
3. Training and inference technologies for foundation models and samples, ensuring data and model asset security;
4. Secure construction, performance optimization, and of confidential training and inference infrastructure for foundation models.Minimum Qualifications:
   • Individuals who are completing or recently completed a PhD in Software Development, Computer Science, Computer Engineering, or a related technical discipline.
   • <

     Preferred Qualifications:

     • Experience in one or more of the following areas: LLMs, multimodal models, AI security, adversarial machine learning, trustworthy AI, or secure systems.
     • Solid foundation in LLM-related algorithms, with hands-on experience in model evaluation, deployment, or safety-related research.
     • Publications in top-tier conferences or journals such as NeurIPS, ICML, ICLR, ACL, EMNLP, NAACL, CVPR, ICCV, or ECCV are a plus.