EngD position on LLM Inference for Knowledge Graph

Artificial intelligence (AI) and knowledge graphs (KGs) have become transformative technologies across numerous domains, enabling advanced decision-making, data integration, and knowledge extraction. AI, particularly large language models (LLMs), excels in natural language understanding and reasoning, while KGs provide structured, interpretable representations of data that enhance the contextual relevance and reliability of AI systems. Their combination paves the way for impactful innovations in various fields, including healthcare, urban planning, environmental monitoring, and public governance.

Research at the intersection of AI/LLMs and KGs is critical for addressing complex, data-driven challenges, especially in land administration, where reliable and interpretable data is paramount. Organizations like the Netherlands’ Cadastre, Land Registry, and Mapping Agency (Kadaster) rely on structured data to maintain legal certainty and provide insights for spatial planning and property rights. Integrating KGs and LLMs can enhance the utility of Kadaster’s datasets, enabling more advanced applications such as multi-modal data analysis, (geospatial) reasoning, and the development of trustworthy AI tools for land administration.

Some of the topics the EngD will investigate include:

• User query interpretation
• Knowledge graph prompting
• Dynamic knowledge fusion

About the project
The Netherlands’ Cadastre, Land Registry and Mapping Agency – in short, Kadaster and the University of Twente (UT) have joined forces to operate at the forefront of knowledge about federated data; the goal is to advance the research field and develop methods and techniques to extract, combine, and analyze information from distributed data sources, while accounting for the principles of ethical conduct, scientific integrity, and open science, to benefit the society.

To realize that goal, Kadaster and UT work together on the UTKa Datalab project under the umbrella of the Centre for Security and Digitalisation (CVD), a collaborative knowledge center based in Apeldoorn, uniting educational institutions, businesses, and public organizations to address challenges in security and digital transformation. The project aims to work on trusted federative data infrastructures based on KG technology and simultaneously explore the potential of mutual augmentation of AI (particularly LLMs) and KG for Land Administration applications. The project will be carried out by a team of five researchers (2 PhD candidates and 3 EngD candidates, including this vacancy) under joint supervision by Kadaster and UT.

The supervision team includes the Geo-Information Processing (GIP) department from UT Faculty ITC, Semantics, Cybersecurity & Services (SCS) department from UT Faculty EEMCS, and the Kadaster Data Science team.

About Kadaster
Kadaster collects and registers administrative and spatial data on property, rights, and assets such as ships, aircraft, and telecom networks, ensuring legal certainty. As the responsible body for national mapping, maintaining the national reference coordinate system, and advising on land use and spatial data infrastructures, Kadaster provides information primarily through online services to civil-law notaries, local authorities, businesses, financial institutions, and individuals. As the custodian of the Key Registers Cadastre and Topography, Kadaster performs its public tasks transparently and in service of society. For further information about Kadaster, please check the following websites: https://www.kadaster.nl/ & https://labs.kadaster.nl/about/.

About GIP
The GIP department at the UT’s Faculty of Geo-Information Science and Earth Observation (ITC) focuses on creating actionable geo-information for diverse stakeholders. GIP addresses critical societal challenges by designing methods to process heterogeneous spatio-temporal data and developing open geo-information solutions. Their interdisciplinary approach combines Geographic Information Science, Remote Sensing, Computer Science, and Digital Humanities, emphasizing co-creation with domain experts to ensure societal relevance and scientific validity.

Requirements:

• An MSc in a related field (e.g., software development, computer science, Geo-Information Science)
• Strong background in at least one of the following areas and interest in the others
• Semantic Web Technologies (e.g., RDF, OWL, SPARQL, GeoSPARQL)
• Natural Language Processing and particularly Large Language Models (e.g., Prompt Engineering, Retrieval Augmented Generation Machine Learning and Deep Learning
• Programming skills (e.g., Python, Java, C#)
• Analytical and problem-solving skills, with the ability to work collaboratively in a multidisciplinary team
• Strong communication skills, with proficiency in English
• Knowledge of Geospatial Information Processing, e.g., working with vector data and spatial database management systems, is a plus

Salary Benefits:

• We offer you a contract for 24 months.
• Competitive salary in accordance with the Dutch Universities’ Collective Labor Agreement (CAO-NU): Gross monthly salary of € 2.872,-;
• Employee status at the UT, including pension and health care benefits. UT is an equal opportunities employer.
• 8% holiday allowance and 8.3% year-end bonus.
• Solid pension scheme.
• A minimum of 41 holiday days for full-time employment.
• Professional and personal development programs through Twente Graduate School.
• Hybrid working opportunities across CVD (1 day per week), Kadaster (1 day per week), and the University of Twente (2 days per week)

For more information about the EngD program at UT, check the UT EngD information page

40 hours per week

Drienerlolaan 5