Technical Stewardship - 6-Month Internship

Job description

Who we are Luxembourg National Data Service (LNDS) is a brand of PNED G.I.E. an organisation created by the Luxembourg Government, to implement Luxembourg’s strategies in research, innovation, and digitalisation. LNDS enables value creation from secondary use of data, for public and private partners and supports the sharing and re-use of public sector data, in a trustable manner. The LNDS service portfolio includes know-how, capabilities, tools, infrastructure, and data services. Through efficient & responsible use of data and improving the secondary use of data, LNDS will support the acceleration of economic, ecological, and societal transitions. Purpose of the job / project You will work closely with LNDS’ Data Stewards, a multidisciplinary team covering topics of Research Data Management, Data Governance, and Applicant support in Data Access Applications. You will contribute to accelerate and facilitate the LNDS role in supporting applicants for Data Access Requests. What you will do You will investigate the further development of an existing Large Language Model (LLM) pipeline based on Retrieval-Augmented Generation (RAG). The objective is to design, implement, and evaluate one or more approaches for schema-to-text and text-to-schema transformations within a defined Data Access Application workflow. The developed system will: Analyze data flow diagrams and populate the corresponding template with accurate and complete information. Create data flow diagrams based on textual requirements and specifications. Verify the internal consistency and logical coherence of data flows or generated outputs. Perform compliance checks against reference materials and knowledge sources provided mainly through the RAG framework. Identify opportunities to improve the quality, completeness, and consistency of data flows or generated outputs. The basic business needs are: To commit to developing an understanding of business goals and operational processes underlying the use case and translating them into effective AI-driven solutions. To evaluate and validate the generated outputs against defined quality criteria and state-of-the-art assessment frameworks. Who you are Master's student in Mathematics, Computer Science, Data Science , or a related field Ability to conduct literature reviews and select relevant publications Critical thinking Team-working ability Technical expectations Understanding of machine learning concepts Familiarity with evaluation methods and metrics Natural Language Processing (NLP) Programming experience, including working with APIs Experience with Open-Source LLMs is a plus Familiarity with RAG concepts Prompt design