PhD AI Security Intern: Safe Agent Protocols & Access

Wayne State University; United States; On-site; USD 41,328 - 68,880; Full time

Job summary

A prestigious university is seeking a PhD Software Engineer Intern to join the AI Security team for winter 2026. The position involves research and development of AI agents focusing on security and privacy. Responsibilities include prototyping, designing risk-based access policies, and collaborating with engineers to integrate solutions into production. Ideal candidates are current PhD students with strong skills in Python and machine learning, looking to apply their knowledge in a real-world setting. This is a unique opportunity to work on cutting-edge security projects in AI.

Qualifications

• Must have at least one semester/quarter of education remaining after internship.
• Strong grounding in LLMs and empirical evaluation.
• Demonstrated ability to translate ideas into working prototypes.

Responsibilities

• Research and prototype identity and attestation for AI agents.
• Design risk-based access policies and scoring.
• Build evaluation harnesses for agent workflows.
• Integrate prototypes into production with engineers.
• Communicate findings through documentation and talks.

Skills

Python

Machine Learning tools (PyTorch or TensorFlow)

Research abilities

Understanding of LLMs/agent frameworks

Problem-solving

Communication skills

Education

Current PhD student in Computer Science, AI/ML, or Security

Job description

A prestigious university is seeking a PhD Software Engineer Intern to join the AI Security team for winter 2026. The position involves research and development of AI agents focusing on security and privacy. Responsibilities include prototyping, designing risk-based access policies, and collaborating with engineers to integrate solutions into production. Ideal candidates are current PhD students with strong skills in Python and machine learning, looking to apply their knowledge in a real-world setting. This is a unique opportunity to work on cutting-edge security projects in AI.