The recent advancements in legal NLP are significantly enhancing the precision and efficiency of legal document processing. Innovations in transformer models, particularly those fine-tuned for legal text, are leading to substantial improvements in tasks such as Legal Entity Recognition (LER) and Cause of Actions (COA) analysis. These models are now incorporating semantic filtering and clustering techniques to better handle the complexities and ambiguities inherent in legal documents. Additionally, the integration of large language models like GreekLegalRoBERTa is expanding the scope of NLP applications in low-resource languages, offering new possibilities for domain-specific tasks. The LegalLens Shared Task highlights the continued need for advancements in detecting legal violations within unstructured text, emphasizing the role of fine-tuned language models in achieving higher accuracy. Notably, the hybrid transformer model with semantic filtering stands out for its innovative approach to improving LER, while the ensemble model for COA similarity analysis offers a novel method for legal research.