International Journal of Science, Engineering and TechnologyAuthors: Gayatri Dekhane, Rinkal Rahane, Gauri Dani, Shruti Sadgir, Sanika Dighe
Abstract: The rising demand for personalized and efficient learning solutions has accelerated the integration of Artificial Intelligence (AI) into modern education. To address this, we present Sahayak, an AI-powered teaching assistant designed to automate study material generation while enabling adaptive learning. The system begins with Optical Character Recognition (OCR) to extract text from handwritten notes, books, and images, ensuring compatibility with both traditional and digital resources. The extracted data is indexed using FAISS, providing efficient knowledge storage and semantic retrieval for large-scale educational content. Leveraging a GPT-based model, Sahayak generates quizzes, flashcards, and worksheets that are contextually aligned with the source material, thereby enhancing learning reinforcement and retention. To support personalization, K-Means clustering groups students based on their performance, enabling adaptive pathways and tailored recommendations that cater to diverse learner needs. The generated content is delivered through teacher-validated dashboards, ensuring accuracy, relevance, and pedagogical trust. This human-in-the-loop approach bridges innovation with reliability, empowering teachers while maintaining academic quality. Experimental results show that Sahayak significantly reduces teacher workload by automating repetitive tasks, while also improving student engagement, comprehension, and outcomes. The system’s design emphasizes scalability and inclusivity, making it adaptable to both resource-rich urban schools and under-resourced rural contexts. By combining automation, personalization, and teacher oversight, Sahayak demonstrates the transformative potential of AI-driven tools in reshaping education. Furthermore, it contributes to the broader vision of accessible, data-driven, and student-centered learning, aligned with global sustainable development goals.
DOI: https://doi.org/10.5281/zenodo.18596444
