Combining advanced numerical methods and cutting-edge computing technologies for imaging of the subsurface and more.
I am an Assistant Professor at KAUST University in the
Physical Science and Engineering division and head of the
Deep Imaging Group
Previously, I worked in Equinor as both a research scientist and reservoir geophysicist. During my time in industry I contributed to the development of geophysical technologies aimed at identifying new discoveries as well as increasing hydrocarbon recovery of existing reservoirs. I have also been involved in the development of several open-source software packages to ease the use of geophysical data and improve reproducibility in the area of inverse problems.
I earned a PhD in Geophysics from the University of Edinburgh as part of the Edinburgh Interferomety Project under the supervision of Prof. Andrew Curtis. My research contributions spanned across the fields of seismic processing, imaging, and inversion through the development of novel methods aimed at using high-order reverberations to improve the quality and resolution of subsurface imaging products.
Distributed, multi-dimensional convolutional operators suited to a variety of seismic processing and imaging applications.
Inclusion of all orders and types of multiples in seismic imaging by solving the Marchenko equations.
Precondition physics-driven inverse problems with non-linear dimensionality reduction networks.
Use of interferometric relations to increase the sensitivity of full-waveform inversion cost functions to target areas.
Utilize multi-measurement acquisition systems to improve processing, imaging, and inversion algorithms.
A Python library aimed at solving large-scale inverse problems in an efficient manner, whilst writing computer code that resembles the underlying mathematica formuation.
A lightweight extension of the PyLops library to compute operators on HPC systems via Dask.
An extension of the PyLops library to compute operators on GPUs. Based on PyTorch, it also allows seamless integration of PyLops operators in PyTorch autograd.
A Python library aimed at optimizing convex, non-smooth functionals by means of proximal operators and solvers.
A Python library collecting most of the Marchenko algorithms available in the literature, all implemented using the same framework and accessible in the same way.
A Python solver for L1 regularized problems. Finds applications in data denoising, compressive sensing, and sparsity promoting inversion in general.
The Deep Imaging Group is constantly looking for highly motivated Ph.D. candidates, Postdoctoral fellows, and Visiting researchers with expertise in one or more of the following areas: