Code Smarter, Not Harder: Uncovering the Energy Footprint of LLM-Generated Code
Organization
Software Engineering Analytics
Abstract
While Large Language Models (LLMs) transform how code is generated, their environmental impact remains unexplored. This thesis dives into the energy efficiency of LLM-generated code compared to traditional human-written code. By profiling and measuring CPU, memory, and power consumption during execution in a specific context(s), we aim to identify inefficiencies in AI-generated code and highlight areas for optimization. We seek to bridge the gap between code automation and sustainable software development, offering insights into how LLMs can speed up coding and make it greener. The exact details of the thesis topic depend on the student's previous knowledge and interests.
Graduation Theses defence year
2024-2025
Supervisor
Hina Anwar
Spoken language (s)
English
Requirements for candidates
Level
Masters
Application of contact
Name
Hina Anwar
Phone
E-mail