I build physics-based models, and the open tools that make them reproducible. My work spans hydrology, soils, agronomy, ecology, ... and music.

• nuee Multivariate analysis for numerical ecology in Python.
• meandre Differentiable hydrology. Process-based catchment models in which every flux and storage term is differentiable, for gradient-based calibration at scale, sensitivity analysis, and hybrid models constrained by conservation laws.
• capyllary Differentiable constitutive laws for unsaturated hydromechanics: coupled flow, poromechanics and contaminant transport in pure Python, runnable in a marimo notebook with nothing to install.

The tangent/suite is an install-free scientific computing in JavaScript, tested against reference tools in R and Python.

• ds Data science: ordination, clustering, classical statistics and machine learning. Companion course: Data Science in the Browser.
• mc Bayesian inference and MCMC sampling.
• sem Structural equation modeling.
• nn Neural networks (CPU, float64).
• Numerical building blocks: proba (probability distributions), grad (automatic differentiation), opt (optimizers), ode (ODE solvers).
• note Local-first computational notebooks running in the browser. No server, works offline.

• jmonlabs Algorithmic music composition with the jmon declarative format (my own Json Musical Object Notation) and jmon/algo, a library for algorithmic composition.
• dustic A lightweight audio player for archive.org collections, installable as a PWA at dustic.app.

• Analyse et modélisation d'agroécosystèmes avec R Course notes on data analysis and modeling for ecologists (source).
• Introduction au calcul numérique avec Python A primer on scientific Python for engineers.

Ph.D. in civil engineering. More at tangent.to.