As software engineering continues to evolve, agile frameworks have become the main paradigm for project management and development. Among these, Scrum is widely adopted by numerous software organizations due to its iterative and incremental development approach, emphasis on team collaboration, and rapid feedback cycles. With the ongoing advancements in artificial intelligence technologies, particularly the recent breakthroughs in large language models, there is growing interest in integrating AI throughout the entire lifecycle to improve efficiency and quality further.
Thesis Type |
|
Student |
Kevin Ha |
Status |
Running |
Presentation room |
Seminar room I5 6202 |
Supervisor(s) |
Stefan Decker |
Advisor(s) |
Yongli Mou Maximilian Kißgen |
Contact |
mou@dbis.rwth-aachen.de kissgen@dbis.rwth-aachen.de |
Thesis Goals
Traditional Scrum teams rely on humans, including the Scrum Master, Product Owner, Developers, and other Stakeholders, to handle requirements analysis, task assignment, coding, testing, and continuous integration. An AI Agent is a software entity capable of perceiving its environment, analyzing information, and autonomously executing tasks. In this thesis, the primary goal of this thesis is to design, implement, and evaluate an AI-empowered agile project management system to empower Scrum teams by embedding AI agents throughout the management and development process. Beyond traditional workflow automation, ScrumiX aims to leverage AI to facilitate backlog management, sprint planning, and risk prediction, as well as AI-driven conversational agents that assist team collaboration. A novel aspect of this system is the facilitation of communication between different AI coding assistants — enabling multiple AI tools to “talk” and coordinate on coding tasks, code reviews, and issue resolutions, thus amplifying the overall efficiency and accuracy of software development.
Research Questions
-
How can AI techniques be effectively integrated into Scrum workflows to support and augment human decision-making without compromising team autonomy?
-
How can multi-agent AI communication be structured to facilitate collaboration and coordination among AI coding assistants within a software development environment?
-
What are the measurable impacts of AI-empowered project management systems on team productivity, communication efficiency, and software quality?
Tasks
-
Conduct a comprehensive literature review on AI applications in agile software development and multi-agent communication systems.
-
Analyze and model Scrum workflows to identify integration points for AI automation and augmentation.
-
Design and implement core AI agent modules for user stories, predictive models for sprint planning, and conversational AI agents.
-
(Optional) Develop a multi-agent communication protocol enabling AI coding assistants to exchange information and collaborate on code-related tasks.
-
Build a prototype of the ScrumiX system incorporating the above modules.
-
Perform case studies and user evaluations with software development teams to gather quantitative and qualitative data on system performance.
-
Refine models and system design based on feedback and evaluation results.
-
Solid understanding of agile software development methodologies, especially Scrum.
-
Background knowledge in machine learning and natural language processing, particularly large language models.
-
Familiarity with multi-agent systems, AI conversational agents, and communication protocols.
-
Experience in software engineering, system design, and web-based application development.
-
Basic knowledge of AI coding assistants (e.g., GitHub Copilot, ChatGPT-based tools) and their integration challenges.