A machine learning approach for continuous development

Complex and ephemeral software requirements, short time-to-market plans and fast changing information technologies have a deep impact on the design of software architectures, especially in Agile/DevOps projects where micro-services are integrated rapidly and incrementally. In this context, the ability to analyze new software requirements and understand very quickly and effectively their impact on the software architecture design becomes quite crucial. In this work we propose a novel and flexible approach for applying machine learning techniques to assist and speed-up the continuous development process, specifically within the mission-critical domain, where requirements are quite difficult to manage. More specifically, we introduce an Intelligent Software Assistant, designed as an open and plug-in based architecture powered by Machine Learning techniques and present a possible instantiation of this architecture in order to prove the viability of our solution.