Learning Management Systems (LMSs) increasingly rely on digital assessments to support automated evaluation, content reuse, and flexible learning scenarios. While the IMS Question and Test Interoperability (QTI) specification provides a standardized, platform-independent format for representing assessment items, its practical adoption remains limited due to fragmented tool support, partial specification coverage, and insufficient integration with execution environments. In addition, assessment content is frequently authored and distributed in document-oriented formats, particularly PDF files, which lack explicit structural and semantic information and are therefore difficult to transform into standardized, machine-processable representations. This paper proposes a hybrid transformation pipeline that combines large language model (LLM)–based document interpretation with a QTI-based metamodel and deterministic model-driven engineering (MDE) techniques to address these challenges. The approach recovers structured assessment items from unstructured documents and reliably transforms them into standardized, LMS-ready representations. Evaluation on real-world repositories, including IMS QTI examples and the Canterbury Question Bank, demonstrates correct semantic preservation, reliable transformation behaviour, and successful import of the generated assessments into an LMS. These findings establish a solid foundation for future extensions to additional LMS platforms.