PhD Position in Secure and Efficient Transformer-Based AI Systems

This fully funded PhD position at Tallinn University of Technology (TalTech) focuses on advancing secure and efficient deployment of Transformer-based AI systems. The project investigates how optimization techniques like quantization, pruning, sparsity, and KV-cache compression impact vulnerabilities to side-channel attacks, fault attacks, hardware Trojans, and soft errors. The candidate will develop lightweight protections for secure and dependable Transformer inference on modern AI accelerators, addressing critical challenges in hardware security and system reliability. The research spans Transformer architectures, AI accelerators, and fault-tolerant design, with a strong emphasis on interdisciplinary collaboration and reproducible experimentation.

• Define research questions and experimental methodologies for secure and reliable efficient Transformer systems.
• Implement and benchmark Transformer models, optimization techniques, and AI accelerator configurations.
• Analyze vulnerabilities to side-channel attacks, fault attacks, hardware Trojans, and soft errors across model components.
• Develop lightweight protection, detection, and mitigation mechanisms for secure and efficient inference.
• Design reproducible experiments and document results for peer-reviewed publications and conference presentations.
• Collaborate with supervisors and TalTech researchers, participate in scientific discussions, and assist in teaching and lab activities.