Advanced Techniques for Evaluating and Auditing AI Systems (with Examples)
ISACA Athens Chapter, in collaboration with code4thought, is organizing the online event “Advanced Techniques for Evaluating and Auditing AI Systems”, focusing on modern techniques for the evaluation and auditing of Artificial Intelligence systems.
The presentation will provide an introduction to the ways in which different types of Artificial Intelligence (AI) systems can be audited — including GenAI, Visual AI, and Tabular / Time Series AI — in terms of fairness, accuracy, and explainability. It will then present the technical audit requirements for AI systems under the EU AI Act and ISO/IEC 42001, followed by examples of audits performed on AI-based employee recruitment systems, both within the framework of the NYC Bias Audit Law and based on user-defined criteria.
Key topics to be covered:
- Methods for auditing different types of AI models — GenAI, Visual AI, and Tabular / Time Series AI — in terms of fairness, accuracy, and explainability. Speaker: Petros Stavroulakis
- Analysis of the areas where automated quality control of AI systems can be applied in relation to the EU AI Act and ISO/IEC 42001. Speaker: Petros Stavroulakis
- Example of an automated audit of an AI system used in employee recruitment, based on the NYC Bias Audit Law. Speaker: Romanos Kapsalis
- Example of an automated audit of an AI system used in employee recruitment, based on user-defined criteria. Speaker: Romanos Kapsalis
Speakers
Petros Stavroulakis is a Responsible AI Consultant at code4thought. His work focuses both on corporate consulting and on the development of AI auditing systems at a research and technical level.
He holds a degree in Electronic Engineering from the University of Southampton in the United Kingdom, as well as an MSc and a PhD from the same university. He has worked as a postdoctoral researcher at the University of Nottingham and at the Foundation for Research and Technology – Hellas in Heraklion, Crete. He has also worked as a Higher Research Scientist at the National Physical Laboratory in London, as well as in multiple companies as Research Manager and Senior Research Manager in the United Kingdom, Germany, and China. Since 2025, he has also been teaching at the Hellenic Army Academy, in the Department of Informatics.
Romanos Kapsalis is an AI Engineer at code4thought, specializing in the evaluation, auditing, and quality assurance of AI systems. His work focuses on the development of techniques and tools for assessing the performance, reliability, and compliance of machine learning models, including bias and explainability audits.
He actively contributes to the development of the iQ4AI platform, supporting the design and implementation of features that enable the internal audit and governance of Artificial Intelligence systems. He also has experience in technical evaluation and AI due diligence projects, helping organizations understand and manage risks related to the adoption of Artificial Intelligence.
He began his career as a Machine Learning Engineer in European research projects at the Institute for Bio-Economy and Agri-Technology of CERTH, where he developed deep learning and machine learning models. He is a graduate of the Department of Computer Engineering and Informatics of the University of Patras, holding an Integrated MSc. His main interests include the development of trustworthy AI systems and modern applications of Large Language Models (LLMs).
In cooperation with the Hellenic American Union.