CRANE

Edith MacArthur, IT Principal Security Architect

Edith MacArthur is an IT Principal Security Architect at a multi-national healthcare insurance company, with over two decades of experience spanning enterprise security architecture, cloud platforms, and information governance across highly regulated sectors. She works closely with architecture, engineering, and product teams to embed security into the full technology lifecycle, from early design through deployment, with a particular focus on AI enabled and data intensive systems.

Edith is a recognised practitioner in secure by design architecture, with deep expertise in AI and machine learning security, cloud security, DevSecOps, and regulatory aligned risk management. Her work focuses on translating business strategy and innovation into practical, scalable security architectures that address confidentiality, integrity, availability, and privacy risks.

Previously, she led security architecture for global CRM platforms and held senior information protection and governance roles across both public and private sectors.

• LinkedIn

Fabio Pierazzi, Associate Professor, University College London (UCL)

Fabio Pierazzi is an Associate Professor in the Department of Computer Science at University College London (UCL), where he is a member of the Information Security Research Group. His research vision is to make AI safe and trustworthy from a systems security perspective, with emphasis on hardening learning-based detection systems operating under adversarial pressure in highly non-stationary contexts. He cares deeply about the practicality of these solutions and regularly engages with industry on the implications of deploying AI in security-critical scenarios.

Together with his Ph.D. students, postdocs, and close collaborators, he has published his research in venues such as USENIX Security, IEEE S&P, AAAI, IEEE SaTML, and ACM TOPS, including work on adversarial training for robust malware detectors, trustworthy dataset curation guidelines that mitigate spatio-temporal biases, knowledge-guided ML to improve generalization against diverse threats, and reinforcement learning that sustains accuracy under concept drift. He also co-authored “Dos and Don’ts of Machine Learning in Computer Security,” which received a Distinguished Paper Award at USENIX Security 2022.

Looking ahead, he is interested in how small language models might be effectively deployed in agentic security workflows, including supporting analysts in multi-step investigations and characterizing the new attack surfaces they may introduce.

• Link