Arbetsbeskrivning

Örebro University and the School of Science and Technology are looking for a doctoral student (PhD student) for the doctoral programme in Computer Science, concluding with a doctoral degree.   Start date: 1st of May 2026 Project description   The multi-disciplinary centre for cyber resilient AI, RESIST, is a national effort funded by the Swedish Strategic Research Foundation (SSF) to bring together leading researchers in AI and cybersecurity to develop novel solutions to cyber resilient AI for the benefit of Swedish industry and society. The vision is to make Sweden a role model in secure trustworthy AI by pioneering cyber resilience across the AI lifecycle. The research program focuses on four key themes: Trustworthy and Verifiable AI, Runtime Security Assurance, Robust and Secure AI-Supported Development, and Resilient Distributed and Agentic AI. RESIST will drive world-class research in the intersection between AI and cybersecurity through a strong, stimulating and well connected international research environment. Research outcomes will be validated in realworld scenarios with industry and public-sector partners. RESIST will also serve as a national hub for cyber resilient AI, promoting education, knowledge sharing, and policy development. This PhD position is associated with the research theme Runtime Security Assurance.   Once trained, AI models can be deployed in a variety of environments, ranging from public repositories such as Hugging Face, to proprietary cloud-based platforms, following the Model-as-a-Service (MaaS) paradigm. These models, whether convolutional neural networks (CNNs), recurrent neural networks, large language models (LLMs), or large reasoning models (LRMs), are designed to respond at inference phase to user-provided inputs with meaningful outputs. However, this significantly broadens the attack surface, increasing the risk of inference-time threats that exploit model interactions with external queries. These threats can undermine robustness, compromise privacy, or circumvent alignment safeguards.   In this project, the PhD student will investigate the protection of AI models at runtime, addressing multiple stages of the inference pipeline. The research will focus on developing techniques to ensure both output reliability under adversarial input attacks and dependable model behaviour in the presence of malicious or policy-subverting prompts. In particular, the project will study mechanisms to defend large language models (LLMs) and large reasoning models (LRMs) against jailbreak attacks and other forms of behavioural manipulation. In addition, the project will develop methods to mitigate inference-time privacy leakage and unauthorized model replication. The overall objective of the thesis is to integrate these techniques into a unified framework for runtime security assurance of AI models. The research is tightly integrated with other research themes within the centre.   Supervision: The doctoral student will be supervised by Professor Mauro Conti (primary) and Dr. Alberto Giaretta (secondary).   The programme, doctoral studentship, entry requirements and selection  To see the job advertisement in its entirety visit: https://www.oru.se/english/career/available-positions/job/?jid=20260051 Information   For more information about the programme and the doctoral studentship, please contact Dr. Alberto Giaretta, email: [email protected] and/or Prof. Mauro Conti, email: [email protected]. For administration issues, contact Head of Unit Martin Magnusson, email: [email protected]. At Örebro University, we expect each member of staff to be open to development and change; take responsibility for their work and performance; demonstrate a keen interest in collaboration and contribute to development; as well as to show respect for others by adopting a constructive and professional approach.   Örebro University actively pursues equal opportunities and gender equality as well as a work environment characterised by openness, trust and respect. We value the qualities that diversity adds to our operations.   Application to the programme and for the doctoral studentship   The application is made online. Click the button “Apply” to begin the application procedure.   For the application to be complete, the following electronic documents must be included:   • CV   • Proof that you meet the general and specific entry requirements (copies of the original certificate and official transcript for bachelor's and master's degrees) • Independent project (degree project)   • Other relevant documents, course and degree certificates verifying eligibility   • Description of research interests - explaining why you are interested in this project and why you would be a good candidate for the role (1 page) As a main rule, application documents and attachments are to be written in Swedish, Danish, Norwegian, or English. Certificates and documents in other languages verifying your qualifications and experience must be translated by an authorised translator to Swedish or English. A list of authorised translators can be obtained from Kammarkollegiet (the Legal, Financial and Administrative Services Agency), www.kammarkollegiet.se/engelska/start.   When you apply for admission, you automatically also apply for a doctoral studentship.   More information for applicants will be found on our career site:  https://www.oru.se/english/career/available-positions/applicants-and-external-experts/   The application deadline is 1st of April, 2026. We look forward to receiving your application!   As we have already made our choices in terms of external collaboration partners and marketing efforts for this recruitment process, we decline any contact with recruitment agencies and advertisers.    As directed by the National Archives of Sweden (Riksarkivet), we are required to deposit one file copy of the application documents, excluding publications, for a period of two years after the appointment decision has gained legal force.