Tristan Coignion

Tristan Coignion is a researcher at University of Lille affiliated with Inria (French National Institute for Research in Digital Science and Technology), actively contributing to software engineering and artificial intelligence research. His work bridges theoretical AI advancements with practical software development challenges, particularly through conference publications at ASE and EASE.

His research interests center on Large Language Models in programming contexts, with specific focus on code optimization efficiency, environmental impacts of AI-generated code, and empirical performance validation. Coignion investigates critical trade-offs between computational speed and energy consumption in LLM-optimized code, while also analyzing real-world effectiveness of AI-generated solutions through platforms like Leetcode.

Recent publications reveal two interconnected research thrusts: the 2025 ASE paper exposes hidden energy costs in LLM-based optimization, challenging assumptions about computational efficiency, while the 2024 EASE study establishes empirical baselines for LLM code performance in competitive programming environments. Together, these works form a cohesive investigation into the sustainability and practicality of AI-assisted software development, highlighting previously overlooked environmental dimensions in the field.