Hybrid LLM + rule-based framework for generating SecBPMN2 security annotations from BPMN models and natural-language requirements, with higher precision and substantially faster annotation than manual expert labeling.

Group-equivariant deep clustering for chest X-rays with faster convergence and clearer clusters, without data augmentation.

Presented at the Twelfth European Big Data Management & Analytics Summer School (eBISS 2024), University of Padova, Italy (July 2024); this work reviews KG-RAG architectures, applications, and open challenges.

Undergraduate thesis on CNN-based brain tumor classification, comparing VGG-16, Xception, ResNet-50, and Inception-V3 on MRI data.