Titulo Estágio
Application of Generative AI to Software Engineering
Áreas de especialidade
Engenharia de Software
Sistemas Inteligentes
Local do Estágio
DEI-FCTUC
Enquadramento
The advent of generative AI (GenAI) models has profoundly reshaped the landscape of various domains, particularly in the software industry.
While the potential benefits of integrating GenAI into the Software Development Lifecycle (SDLC) are widely acknowledged, including accelerated development, improved code quality, and enhanced developer productivity, a focused research effort is crucial to understand and optimize its application, particularly within smaller-scale projects or specific organizational contexts. Such a research endeavor could provide valuable insights into practical implementation strategies, address potential challenges, and inform best practices for maximizing the utility of GenAI in real-world software engineering scenarios.
Objetivo
The expected outcomes of such a research work are multifaceted and impactful. As a minimum, we expect the student to deliver a white-paper presenting the State-of-Art of GenAI for software development.
Quantitatively, the research must provide evidence on the specific improvements in development speed, reduction in errors, or increase in productivity achieved through GenAI adoption.
Qualitatively, it should shed light on the usability, user satisfaction, and the challenges faced by developers when interacting with GenAI tools, including issues like "hallucinations," security concerns, and ethical considerations.
The expected outcome includes the development of recommendations for optimal prompt engineering, strategies for human-AI collaboration, and guidelines for integrating GenAI into existing development workflows.
Ultimately, this research work shall provide the student with a thorough understanding of how generative AI can be effectively leveraged to enhance software development practices. This competence shall be a most valuable asset in the search for a relevant position in the modern software industry.
Plano de Trabalhos - Semestre 1
The proposed work shall follow a structured approach, encompassing several key steps.
In the first semester it will involve a thorough literature review to identify existing research, tools, and established use cases of GenAI in various SDLC phases, such as requirements analysis, design, code generation, testing, documentation, and maintenance.
The sources for this research are abundant and diverse. For academic literature, prominent databases like IEEE Xplore, ACM Digital Library, SpringerLink, and arXiv would be invaluable. These platforms host numerous papers on AI in software engineering, large language models, and their applications. Beyond academic sources, industry reports from leading technology firms (e.g., Microsoft, Google, AWS, GitHub, PwC, McKinsey) offer practical insights and case studies of GenAI adoption. Open-source GenAI tools and platforms (e.g., GitHub Copilot, OpenAI's API, Hugging Face) provide direct avenues for experimentation. Furthermore, we expect the student to conduct interviews or surveys with software developers, team leads, and project managers in organizations that have adopted or are considering GenAI tools (mostly in the Coimbra software development community) shall offer rich qualitative data on their experiences, challenges, and perceived benefits.
Plano de Trabalhos - Semestre 2
In the second semester the core of the effort shall involve hands-on work, such as controlled experiments or case studies using ongoing development projects with industry partners. This may involve comparing development efficiency and code quality in projects using GenAI tools versus those relying solely on traditional methods, or analyzing the impact of GenAI on specific developer tasks like bug fixing or test case generation. Data collection shall involve metrics like lines of code generated, time taken for specific tasks, number of bugs identified, and qualitative feedback from developers.
Condições
The student shall have access to b-on (the bibliographic references source).
A work place in the Software and Systems Engineering (SSE) research group laboratory.
Remote access to a machine with a modern GPU with significant computing power.
Observações
There is no scholarship funding available for this MSc dissertation work.
Orientador
Mário Alberto da Costa Zenha Rela
mzrela@dei.uc.pt 📩