If you want to get in touch with me about something or just to say hi, send me an email.
Mathematics for Natural Sciences
Science-based data science and AI consulting for agriculture, ecology, geology and environmental research
Services
Predictive Modeling & Workflows
Advanced statistical models and machine learning pipelines for investigating patterns and predicting outcomes in living systems. From investigation with advanced biostatistics to predictions with machine learning, we develop computational workflows that transform complex ecological and agricultural data into science-based decision making processes.
Machine learning, dose-response modeling, time series analysis, neural networks
Physics-Based Modeling
Mechanistic models grounded in physical laws for natural systems such as water, soil, and the subsurface. Where the data allow, we couple these process-based models with data-driven components and differentiable programming, so they can be calibrated against observations and quantify their own uncertainty while remaining physically interpretable rather than black boxes.
Process-based modeling, differentiable simulation, inverse problems, data assimilation, uncertainty quantification
Reproducible Production Environments
Bringing scientific models and analyses from research code to robust, reproducible production environments. The work centres on computational reproducibility: controlled environments with pinned dependencies, full provenance of data and results, and automated, resource-aware execution, so that analyses run reliably and can be reproduced exactly rather than remaining one-off scripts.
MLOps, computational reproducibility, provenance, dependency control, automated execution
Recent projects
See all projects
Corn Nitrogen Response Modeling
Dose–response modeling of corn nitrogen fertilization, pairing canonical agronomic forms with modern machine learning and calibrated uncertainty
Python PyTorch GPyTorch XGBoost
+2
Goblin
Physics-based mineral prospectivity model: geophysical inversion, 3D subsurface reconstruction by diffusion, and at-depth uncertainty (work in progress)
PyTorch Diffusion models Geophysical inversion Geological modeling
+1
`meandre` - Differentiable Hydrology
Process-based hydrological modeling with automatic differentiation for gradient-based calibration and hybrid physics-AI models
PyTorch DuckDB Hydrology