The growing emphasis on personalized learning in educational settings necessitates innovative tools and techniques to enhance student engagement and comprehension. This paper focuses on automating the generation of personalized questions tailored to individual learners’ needs, particularly within Stage 1 of a multi-phase framework. By leveraging Natural Language Processing and Deep Learning methodologies, we propose a novel system that aligns question difficulty and content with a learner’s proficiency and preferences. The study employs a dataset comprising domain-specific content to fine-tune transformer-based models for generating contextualized and adaptive questions. Key findings from Stage 1 indicate the efficacy of the proposed framework in dynamically creating diverse and pedagogically relevant questions, setting a foundation for comprehensive personalization in subsequent stages. This study contributes to the evolving field of intelligent tutoring systems, aiming to bridge gaps in traditional assessment paradigms through automation and learner-centric design.