Evidence-Based AI Safety, Assurance and Evaluation for MSCA Doctoral Networks

I am Dr Abdul Razzaq, Assistant Professor in the Department of Computer Science and Information Systems at the University of Limerick, Ireland. I am interested in contributing to an MSCA Doctoral Network as a University of Limerick beneficiary representative, doctoral supervisor, work-package leader, trainer, and research collaborator.

My expertise lies at the intersection of software engineering, evidence-based AI safety, trustworthy and agentic AI, AI assurance, evaluation, explainability, governance, and dependable software-intensive systems.

I can contribute to consortia developing doctoral research and training in areas such as:

• evidence traceability for AI safety, reliability, and performance claims;
• assurance cases for autonomous, adaptive, and agentic AI systems;
• evaluation and benchmarking of reasoning, planning, and tool-using agents;
• robustness, uncertainty, failure-mode, and unintended-behaviour analysis;
• software engineering methods for dependable AI-enabled systems;
• human oversight, intervention, accountability, and responsibility allocation;
• explainability, provenance, reproducibility, and decision traceability;
• AI governance, validation, and alignment with European regulatory and assurance requirements;
• empirical evaluation of AI systems in realistic application environments.

A potential University of Limerick contribution could include leadership of a work package provisionally titled:

“Evidence, Evaluation and Assurance for Trustworthy AI Systems.”

This work package could establish common evaluation methods, evidence requirements, validation protocols, and assurance mechanisms across the network’s doctoral projects.

A potential doctoral research project could investigate:

“Evidence-Based Assurance of Adaptive and Agentic AI Systems.”

The project could examine how evidence about system behaviour, robustness, uncertainty, human oversight, and safety can be generated, connected, evaluated, and maintained throughout the AI-system lifecycle.

I can also contribute network-wide doctoral training in:

• trustworthy and responsible AI;
• software and AI assurance;
• empirical research methodology;
• evaluation and benchmark design;
• reproducible research;
• evidence-based safety assessment;
• academic writing and research communication.

I am particularly interested in Doctoral Networks addressing trustworthy AI, AI safety, responsible agentic AI, human–AI collaboration, autonomous systems, cyber-physical systems, healthcare technologies, critical infrastructure, and other high-impact software-intensive domains.

The University of Limerick could potentially contribute doctoral supervision, interdisciplinary collaboration, specialised training, research activities, and secondment opportunities, subject to the relevant institutional approvals and consortium arrangements.

I welcome contact from coordinators and consortium members who are developing an MSCA Doctoral Network and are seeking a beneficiary or work-package leader with expertise in evidence, evaluation, software assurance, and trustworthy AI.