Research

Built an interpretable multiclass model to predict DPE energy labels (A–G) from open ADEME data, with evaluation focused on class imbalance, robustness, and next-step improvements.

Human-Centered research initiative exploring intergenerational disconnection through qualitative inquiry and participatory design.

Alert-aware electricity-demand forecasting + DPE risk diagnostics, packaged as a decision-support tool that recommends concrete peak-reduction actions during Ecowatt stress windows.

Exploration of real-time data processing techniques and analysis pipelines for streaming datasets.