With rapid advances in artificial intelligence, especially large language models (e.g., OpenAI GPT, Anthropic Claude, Google Gemini), AI has demonstrated great potential in natural language processing, code generation, and automated testing. Programming, traditionally a highly specialized and creative task, is increasingly supported and partially automated by AI. Tools such as GitHub Copilot and OpenAI Codex already generate code snippets based on natural language prompts, boosting developer productivity. However, current AI code generation systems face challenges such as limited context understanding, inconsistent code quality, lack of adaptation to complex software engineering workflows, and limited capacity for continuous interaction and collaboration. Therefore, building an intelligent coding agent system capable of autonomous planning, requirement comprehension, code writing, debugging, and optimization has become a cutting-edge research topic in academia and industry.
Thesis Type |
|
Status |
Open |
Presentation room |
Seminar room I5 6202 |
Supervisor(s) |
Stefan Decker |
Advisor(s) |
Yongli Mou |
Contact |
mou@dbis.rwth-aachen.de |
This thesis aims to design and implement a Coding AI Agent System based on large language models and multi-agent architectures to achieve the following:
-
Develop an LLM-based Coding AI Agent System for supporting the full programming loop.
-
Investigate natural language-driven requirement parsing and task decomposition methods.
-
Integrate code generation, code review, automated testing, and debugging capabilities to ensure code quality.
-
Explore interaction and collaboration mechanisms between AI agents and human developers to improve system usability.
- Design cost-effective strategies for controlling computational resources and operational costs.
-
Evaluate the system’s performance and efficiency gains on typical programming tasks.
To achieve the above goals, this thesis will focus on answering the following research questions:
-
RQ1: How to design effective task decomposition and scheduling mechanisms that enable the Coding AI Agent to plan and execute programming workflows efficiently?
-
RQ2: How to enhance code generation modules with improved context understanding and quality assurance, integrating static analysis and multi-turn feedback?
-
RQ3: How to enable seamless and effective interaction between AI agents and human developers to support dynamic requirement changes and collaborative programming?
-
RQ4: How to control computational and operational costs in the Coding AI Agent System?
-
A solid understanding of deep learning, especially large language models.
-
Proficiency in software engineering and programming, including experience with common programming languages and version control systems like Git.
-
Familiarity with multi-agent systems, task coordination, and distributed artificial intelligence.
-
Knowledge of code generation techniques and automated testing methodologies, static code analysis, and debugging strategies.
-
Research skills in literature review, experimental design, data analysis, and performance evaluation.