EU remote
Thesis Work - Generative AI for Scenario and Injection Design in IT/OT Cybersecurity Exercises
About this role
Are you a student looking to apply your knowledge to real-world solutions that support the energy transition? With us, you will have the opportunity to complete your thesis project close to the business, working alongside experienced specialists and tackling challenges where your ideas and perspectives can make a real difference. About the Thesis Work The Thesis will be held under Corporate Security & Resilience function at Vattenfall which is a modern, flexible, and agile unit that is operating in a challenging and rapidly changing environment.
We strive to embed security & resilience into all our business operations and we apply a holistic and risk-based approach to constantly adopt and improve our security posture. Background and Scope Cybersecurity exercises, including tabletop exercises and functional exercises are an important part of organizational resilience, particularly for critical infrastructure operators where information technology (IT) and operational technology (OT) environments are interconnected.
The effectiveness of such exercises depends on the realism, technical accuracy, and relevance of the scenario and its supporting injects (alerts, emails, media reports, decision points). However, developing scenarios and injects is often manual and resource intensive activity. Exercise planners must translate and combine multiple variables into a sequence that is credible for both IT and OT participants. Generative AI and large language models can support on scenario development, structured content generation and role-based dialogues, while agentic capabilities can enable more adaptive exercises by proposing subsequent injects.
The thesis will investigate how generative AI can support both the design and adaptive facilitation of IT/OT cybersecurity exercises. The thesis will follow a qualitative approach, focusing on the development of a conceptual framework and workflow specification for AI assisted exercise design. The proposed framework and workflow will be evaluated through expert review and walkthroughs with focus on real world applicability and relevance for IT/OT cybersecurity environments.
The thesis is expected to deliver: An academic framework for generative AI assisted scenarios and injection design; A small library of reviewed example artifacts demonstrating the framework; A documented workflow covering source preparation, scenario generation, state tracking, adaptive inject selection, human approval and debrief support. A limited demonstrator or simulation showing how participant decisions can change subsequent injects and scenario outcomes.