LLMs in the Loop: AI to Build Python Tools for Advanced Transport Data

Overview

Learn how transport professionals can use AI and Python together to build smarter, faster, and more reliable analysis tools.

Date: Wednesday, 24 June 2026
Time: 12:00 PM – 2:00 PM AEST
Format: Online Webinar

Large language models (LLMs) are transforming how transport engineers and analysts write code, but their effectiveness depends on how they are guided.

This practical, demonstration-driven webinar introduces participants to using LLMs as a development partner for building Python-based tools in transport data analysis. Through live examples, attendees will see how LLMs can accelerate the development of demand models, data pipelines, and analytical workflows, while also learning how expert oversight is essential to producing valid and reliable outcomes.

The session highlights an expert-in-the-loop approach, where transport domain knowledge remains the critical ingredient that keeps AI-assisted development accurate, statistically rigorous, and practically useful.

Participants will explore how prompt engineering, model validation, and transport-specific framing can significantly improve the quality of LLM-generated code for applications such as demand forecasting, mode choice modelling, and network analysis.

Learning Objectives

By the end of this webinar, participants will be able to:

• Explain how LLMs can be used as coding assistants in transport modelling and data analysis workflows
• Identify common failure modes of LLM-generated code in technical transport contexts, including statistically incorrect implementations and misspecified models
• Apply an expert-in-the-loop methodology to guide LLM output toward valid, domain-appropriate solutions
• Use prompt engineering strategies tailored to transport modelling tasks such as mode choice, demand forecasting, and network analysis
• Critically evaluate LLM-generated code for correctness against transport theory and statistical rigour

Key Takeaways

• LLMs can dramatically accelerate Python tool development for transport analysis when guided by practitioners with strong domain knowledge
• Without expert oversight, LLMs may produce code that is syntactically correct but analytically flawed, including incorrect likelihood functions, invalid assumptions, or misspecified models
• A structured human-in-the-loop workflow is essential for reliable AI-assisted development
• Prompt design matters — transport-specific terminology and modelling concepts produce substantially better outputs than generic prompts
• LLM-assisted development should be viewed as a force multiplier for experts, not a replacement for expertise

Who Should Attend

This webinar is designed for:

• Transport planners and engineers working with demand modelling, network analysis, or large-scale travel survey data
• Transport researchers applying econometric or statistical methods, including discrete choice modelling, regression, and simulation
• Data analysts and modellers in public agencies, consultancies, or academia seeking to integrate LLMs into Python workflows
• Practitioners interested in understanding where AI-assisted coding helps — and where it breaks down — in technical transport applications
• Anyone exploring the responsible use of generative AI in engineering and applied science