Projects & Software
My own open-source work, alongside projects from my research group that I’m glad to put in the spotlight.
GEARS is an open-source Python package for simulating, forecasting, and optimising electric vehicle charging demand at any scale — from a single site to a national fleet.
Gaussian-mixture and VAE-based session generation
Forecasting with SARIMA, NHiTS, and PatchTST
A V1G smart-charging optimiser
Benchmarked against 11 public EV charging datasets
Full CI/CD via GitHub Actions, 300+ tests
I’m the main contributor, working with Bachir Hamrouche. The package is moving to EDF Lab’s open-source GitHub organisation.
GraphToolbox
This is Eloi Campagne‘s project — he wrote essentially all of it as part of his PhD. I’m featuring it here to promote his work.
GraphToolbox is a Python package for graph machine learning focused on time-series forecasting, with tools for data handling, model building, training, evaluation, and visualisation.
Data handling and preprocessing for graph datasets
GNN models: GCN, GraphSAGE, GAT, and 50+ other layer types
Training and evaluation utilities for graph-based models
Visualisation tools for graph data and results
Runs on CPU, CUDA, or Apple MPS, auto-selected at runtime
Built by Eloi Campagne, with contributions from Itai Zehavi. Currently on Eloi’s own GitHub; moving to EDF’s open-source organisation soon.