Context

Machine failures in complex industrial systems can lead to significant downtime and costly service interventions. When a failure occurs, identifying not just what failed but why it failed remains a slow, manual process heavily reliant on expert intuition. The challenge is compounded by intricate failure mechanisms, diverse operating conditions, and data that is fragmented across sensor streams, maintenance logs, and error records.

Traditional root cause analysis are either too subjective or too resource-intensive for routine use, and they frequently miss novel or complex failure patterns. Emerging causal AI techniques offer a promising alternative: by modelling directed causal relationships from data rather than relying on correlations alone, they can uncover why failures occur and support more targeted corrective actions. However, applying these methods to real industrial systems remains difficult due to data heterogeneity, limited failure examples, and the need to incorporate engineering knowledge alongside learned structure.

Project

The central research challenge of this PhD position is to automatically learn why failures occur from heterogeneous industrial data - multivariate time-series sensor data, maintenance logs and system knowledge included in engineering models and manuals - and to do so reliably under the practical constraints of limited failure examples and continuously evolving operating conditions. You will design hybrid causal discovery methods (with a focus on time series) that combine complementary algorithmic families, including constraint-based, Bayesian, and gradient-based approaches, to extract causal graphs that capture the true dependencies driving failures. These data-driven graphs will be integrated with causal structure derived from physics-based models through a merging strategy that reconciles the two sources into a unified representation. Rather than treating this as a one-off analysis, the causal graph will be kept current through online updating strategies that incorporate incoming operational data as new observations arrive. Once a reliable causal graph is in place, the focus shifts to inference: estimating the actual effect sizes of candidate root causes using machine learning-based techniques such as Double Machine Learning, Meta-Learning, and Causal Forests, which are capable of handling complex, nonlinear relationships.

This PhD position is part of the CausAICA project, a collaborative research project supported by Flanders Make, the strategic reserach center of the manufacturing industry. The project aims at developing a causal AI framework for automated root cause analysis of failures in complex industrial machinery. The project integrates causal discovery and inference, physics-informed modelling, and agentic AI to build a system that links sensor data, maintenance logs, and engineering knowledge into explainable failure diagnoses. This PhD is embedded in the methodological heart of the project, responsible for developing the core causal discovery and inference methods that underpin the framework.

The methods developed throughout the PhD will be validated on realistic research demonstrators, as well as real-world industrial use cases leveraging operational field data provided by the industrial partners in the project.

We are seeking a highly motivated, enthusiastic, passionate, and communicative researcher, with a proactive and creative attitude who is eager to explore innovative solutions. If you recognize yourself in the story below, then you have the profile that fits the project and the research group:

• I have a Master's degree in Computer Science, Artificial Intelligence, Electrical Engineering, Mathematical Engineering or a related field and performed well above average in comparison to my peers.
• I am proficient in written and spoken English.
• During my courses or prior professional activities, I have gathered experience with machine/deep learning, and can demonstrate a strong affinity with these fields. Prior experience with causal machine learning, time series analysis and/or agentic AI is a plus.
• I am proficient in Python and am familiar with data science and machine/deep learning toolkits.
• As a PhD researcher at KU Leuven, I perform research in a structured and scientifically sound manner. I read technical papers, understand the nuances between different theories and implement and improve methodologies myself.
• Based on interactions and discussions with my supervisors and the colleagues in my team, I set up and update a plan of approach for the upcoming 1 to 3 months to work towards my research goals. I work with a sufficient degree of independence to follow my plan and achieve the goals. I indicate timely when deviations of the plan are required, if goals cannot be met or if I want to discuss intermediate results or issues.
• In frequent reporting, varying between weekly to monthly, I show the results that I have obtained and I give a well-founded interpretation of those results. I iterate on my work and my approach based on the feedback of my supervisors which steer the direction of my research.
• I feel comfortable to work as a team member and I am eager to share my results to inspire and being inspired by my colleagues.
• I value being part of a research group which is well connected to the mechatronics industry and I am eager to learn how academic research can be linked to industrial innovation roadmaps.
• During my PhD I want to grow towards representing the research group on project meetings or conferences. I see these events as an occasion to disseminate my work to an audience of international experts and research colleagues, and to learn about the larger context of my research and the research project.

We encourage candidates from diverse backgrounds and experiences to apply, as we believe that different perspectives contribute to better research and innovation.

Application Instructions for the PhD vacancy

To apply for this position, please use the online application tool and ensure that you submit the following documents in a single PDF file:

1. Motivation Letter: A letter (maximum 1 A4 page) addressing your strengths and qualifications in relation to the project.
2. Complete Academic CV: A detailed CV including information about your education, current position, work experience (if any), employment gaps (if any), interests, extracurricular activities, international experiences, and projects demonstrating your programming/software skills, background knowledge relative to the project and level of expertise.
3. List of Publications: If applicable, provide a list of your publications, including DOIs. Please do not include PDFs of the publications.
4. Copies of Diplomas: Include copies of your BSc and MSc degrees.
5. Transcript of Records: Provide transcripts for your BSc and MSc degrees. If you have not yet completed your Master's degree, include your available credits and scores, as well as a list of courses you are taking in the upcoming semester.
6. English Summary of Master Thesis: A summary of your master thesis in English (maximum 1 A4 page, or 2 pages max when including a figure).
7. Proof of English Language Proficiency: Documentation demonstrating your proficiency in English (TOEFL, IELTS, …), if available.
8. Reference Letter or Contact Details: A reference letter or the contact information for one reference who can provide a recommendation letter upon request.

The position will be hosted within the collaborative and internationally oriented research environment at KU Leuven, one of the world's leading universities (ranked among the top 100 globally). Founded in 1425, KU Leuven has been a center of learning for nearly six centuries and is Belgium’s highest-ranked university, as well as one of the oldest and most renowned universities in Europe. KU Leuven provides a truly international experience, high-quality education, world-class research, and cutting-edge innovation, having topped Reuters' ranking of Europe's most innovative universities for four consecutive years.

We offer:

• A fully funded 4-year PhD scholarship, with a remuneration package competitive with industry standards in Belgium, a country with a high quality of life and excellent health care system.
• Ample occasions to develop yourself in a scientific and/or an industrial direction. Besides opportunities offered by the research group, further doctoral training for PhD candidates is provided in the framework of the KU Leuven Arenberg Doctoral School (https://set.kuleuven.be/phd), known for its strong focus on both future scientists and scientifically trained professionals who will valorise their doctoral expertise and competences in a non-academic context.
• Opportunities to collaborate in groundbreaking research and participate in international conferences.
• Access to state-of-the-art infrastructure and a range of university benefits (health insurance, etc.).
• A dynamic, passionate team of fellow PhD students and test engineers.

As a PhD candidate, you will be based at KU Leuven’s Bruges Campus (https://www.kuleuven.be/english/bruges), as part of a dynamic and interdisciplinary team of AI researchers, with access to state-of-the-art lab facilities to experimentally validate your findings in close collaboration with industrial partners.

The successful candidate will be encouraged to present their research at international conferences and national events, with a strong emphasis on publishing high-quality conference papers and journal articles. They will benefit from our robust international research and industrial network, which is actively involved in this project.

KU Leuven Campus Bruges, located in the magnificent medieval city of Bruges in West Flanders, offers a vibrant academic setting in close proximity to a network of companies. The campus features newly established labs to support both educational and research needs.

For more information please contact Prof. Mathias Verbeke, mail: mathias.verbeke@kuleuven.be.