Civil and Environmental Engineering
Applications for 2025-2026 open 1 July 2025.
Integration of Large-language models in agent-based models
Project code: ENG035
Supervisor(s):
Hyesop Shin
Discipline(s):
Civil and Environmental Engineering
School of Geography
Project
Students will work on designing and implementing simple ABM scenarios where selected agents are driven or assisted by LLMs. Potential applications include disaster response, transportation systems, or social simulations.
Tasks
Tasks may include setting up communication protocols between ABM platforms (e.g. Mesa or NetLogo) and LLMs via APIs, crafting prompt-based behaviour scripts, and evaluating the performance or plausibility of LLM-driven agents compared to rule-based agents.
Skills Required
- Basic programming skills (preferably Python)
- Familiarity with agent-based modelling concepts
- Comfort with using AI tools like ChatGPT for collaborative coding and debugging
Prerequisites
- Introductory knowledge of Python
- Interest in simulation, AI, or computational social science
Timing
- Duration: 3 months (full-time or part-time options possible)
- Ideal for summer research or internship projects (November–February in the Southern Hemisphere)
Ideal student
The project is well suited for students with interdisciplinary interests in computer science, artificial intelligence, and the social sciences.