SECURE LOCAL GENERATIVE AI CODING ASSISTANT: A CDIO APPROACH

Generative Artificial Intelligence, referred to as GEN AI, is a buzzword that has gained
immense popularity in recent years. The progress in generative AI has given rise to many
applications in the domain of education. We have seen many of those models as coding
assistants that enhance students’ learning experience in introductory programming courses
such as Python. In this paper, we present a new tool as a coding assistant that integrates the
strengths of these models in computing education, showing how a CDIO approach can be
employed to develop secure and effective AI-powered coding assistants that support novice
learners while preserving academic integrity and ethics involved in such tools. This paper
discusses the implementation of a containerized solution utilizing generative AI to provide
accessible, secure, and efficient offline development tools. We have developed an AI code
assistant that works offline, ensuring better privacy for students and learners based on the
qwen2.5-coder:0.5b model and features a user-friendly Streamlit interface, all within a
Docker container for offline execution to facilitate easy deployment and scalability and ensure
security. This tool aligns with many CDIO standards to ensure a better educational experience
both for learners and educators, which will be discussed in depth. Afterward, we will present
the ethical implications of the implementation and then present the different components of
the tool, to finally discuss in depth what can be improved for such solutions.