PhD: Secure & Privacy‑Preserving Collaborative LLMs

Aalborg Universitet; Denmark; On-site; DKK 312,000 - 402,000; Full time

Job summary

Aalborg University’s Department of Electronic Systems invites applications for a PhD stipend in secure machine learning within Electronic and Electrical Engineering, starting Nov 1, 2026 or as soon as possible.

The department is renowned for its AI, security, and cyber research, with a strong international environment and collaborations across faculties. The PhD position emphasizes privacy-preserving language models, distributed training, and cryptographic protections.

Qualifications

• Master's degree or equivalent in a relevant field and strong motivation for advanced AI security research.
• Excellent written and verbal English communication skills.
• Ability to work independently and in interdisciplinary teams.
• Prior background in privacy-preserving machine learning or cryptographic protocols is advantageous but not required.

Responsibilities

• Join the cybersecurity group focusing on secure and privacy-preserving collaborative language models.
• Develop frameworks for privacy-preserving collaboration among language-model agents.
• Explore federated/distributed learning with secure aggregation and cryptographic protections.
• Investigate defenses against model extraction and semantic leakage in multi-agent systems.

Skills

Motivation for research in AI and secu

English communication

Independent and collaborative work

Privacy-preserving ML (advantage)

Education

Master's degree in information engineering / computer science / cybersecurity / software engineering

Job description

Aalborg University’s Department of Electronic Systems invites applications for a PhD stipend in secure machine learning within Electronic and Electrical Engineering, starting Nov 1, 2026 or as soon as possible.

The department is renowned for its AI, security, and cyber research, with a strong international environment and collaborations across faculties. The PhD position emphasizes privacy-preserving language models, distributed training, and cryptographic protections.