PNNL

José Luis Hernandez Mejia is a geoscience machine learning (ML) research associate at Pacific Northwest National Laboratory (PNNL), where he develops scientific ML methods to characterize, monitor, and forecast complex subsurface and environmental systems.

His research integrates ML, uncertainty quantification, physics-based simulation, and high-performance computing to address challenges in Earth and energy sciences. His work focuses on inverse modeling, probabilistic prediction, and scientific AI for geoscientific applications. He develops generative AI methods, including Conditional Diffusion Models for uncertainty-aware estimation of hydrogeological parameters from geophysical observations and transformer-based neural networks for forecasting multidimensional environmental and geospatial time series.

Hernandez Mejia earned a PhD in petroleum engineering as well as an MS in statistics and an MS in petroleum engineering from The University of Texas at Austin. His expertise spans scientific ML, inverse problems, uncertainty quantification, geostatistics, optimization, high-performance computing, and subsurface energy systems. His research interests include environmental remediation, geothermal energy, carbon management, Earth system modeling, and the development of AI-driven models, optimization algorithms, and computational workflows for geo-energy systems.

• Generative AI
• Inverse modeling
• Optimization
• Geostatistics
• Surrogate models 
• Scientific ML 
• Statistics

• PhD in petroleum engineering, The University of Texas at Austin
• MS in statistics, The University of Texas at Austin
• MS in petroleum engineering, The University of Texas at Austin
• BS in petroleum engineering, Instituto Politécnico Nacional

• Society for Industrial and Applied Mathematics

• 2020 Society of Petroleum Engineers First Place for International Student Paper
• 2026 Waste Management Symposia “Papers of Note” rating for the paper “Generative AI for Calibrating Subsurface Remediation Simulators Using Time-Lapse Electrical Resistivity Tomography.”   

• Hernandez Mejia, J. L., T. C. Johnson, G. E. Hammond, and P. Jaysaval. 2026. “Conditional diffusion models for hydrogeologic parameter estimation from electrical resistivity tomography time-lapse data.” Advances in Water Resources 105312.