Overview
Modern analytics applied to large, complex subsurface datasets.
Modern geoscience generates volumes of data that traditional workflows struggle to exploit. We apply machine learning, statistical analytics and Python-based automation to geophysical, petrophysical and production datasets, extracting predictions, patterns and insight that conventional interpretation alone cannot deliver, while keeping geological and physical realism at the centre of every model.
- Supervised ML for facies predictionRandom forests, gradient boosting and neural networks for facies classification from well logs and inverted seismic volumes.
- Unsupervised clusteringK-means, Gaussian-mixture and self-organising-map clustering for seismic facies and rock-type identification.
- Rock-physics MLHybrid physics + ML workflows that combine Gassmann / Hashin-Shtrikman bounds with data-driven calibration.
- Subsurface property predictionPorosity, permeability, water saturation and lithology prediction from seismic and well data, with uncertainty bars.
- Petrophysics automationAutomated log QC, multi-well normalisation and pseudo-log generation for portfolio-scale studies.
- Workflow automationPython-based automation of repetitive interpretation, QC, mapping and reporting tasks, including Petrel scripting.
- Large-dataset integrationIngestion, QC and integration of multi-source seismic, well, production and operational datasets into queryable data stores.
- Reproducible & auditable workflowsVersion-controlled, notebook-based workflows so that every result can be audited, reproduced and updated.
Tools & Platforms