1. médialab Sciences Po
  2. Team
  3. Lydia Passet
Lydia Passet profile picture

Lydia Passet

I am exploring how matrix regression can enhance the semantic representations of language models in order to refine the analysis of political discourse.

Research Engineer

A graduate with a dual Master’s degree in Data Science (École Polytechnique / HEC Paris) and in Political Philosophy and Ethics (Sorbonne University), Lydia is interested in the application of machine learning to issues of public interest and algorithmic fairness.

After working on chatbots for the ecological transition at ADEME, and then helping to develop open-source software at Linagora, she joined the médialab in April 2026 for a research placement with Alex Kindel.

She applies matrix representation methods to speeches made in the US Congress in order to study the evolution of political stances on immigration.