Large Language Models (LLMs) have recently demonstrated strong capabilities in natural language understanding and generation, enabling new forms of interaction between users and data systems. Among these applications, text-to-SQL generation represents a particularly relevant use case in enterprise environments, where relational databases remain the backbone of data storage and business analytics. However, directly applying general-purpose LLMs to complex business databases poses significant challenges, including schema complexity, SQL dialect constraints, domain-specific business logic, and cost-efficiency considerations. This thesis investigates the effectiveness of different fine-tuning strategies for adapting LLMs to text-to-SQL tasks in enterprise-like settings. In particular, Supervised Fine-Tuning (SFT) and Contextual Fine-Tuning (Context FT) are systematically compared across multiple configurations.