Matteo Ravasi

Assistant Professor in Computational Geophysics, KAUST University, Saudi Arabia

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.

Research Activities

Large-scale inverse problems in geophysical processing

Distributed, multi-dimensional convolutional operators suited to a variety of seismic processing and imaging applications.


Full-wavefield redatuming and imaging

Inclusion of all orders and types of multiples in seismic imaging by solving the Marchenko equations.


Deep-learning for seismic inverse problems

Precondition physics-driven inverse problems with non-linear dimensionality reduction networks.


Novel objective functions for target-oriented waveform inversion

Use of interferometric relations to increase the sensitivity of full-waveform inversion cost functions to target areas.


Multi-measurements in seismic data for processing and imaging

Utilize multi-measurement acquisition systems to improve processing, imaging, and inversion algorithms.




- A Linear-Operator Library for Python

A Python library aimed at solving large-scale inverse problems in an efficient manner, whilst writing computer code that resembles the underlying mathematica formuation.


- Distributed Linear Operators

A lightweight extension of the PyLops library to compute operators on HPC systems via Dask.


- Linear operators on GPUs

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.


- Proximal Operators and Solvers in Python

A Python library aimed at optimizing convex, non-smooth functionals by means of proximal operators and solvers.


- A bag of Marchenko algorithms implemented on top of PyLops

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.


- Spectral Projected Gradient for L1 minimization

A Python solver for L1 regularized problems. Finds applications in data denoising, compressive sensing, and sparsity promoting inversion in general.



Machine Learning in Geoscience

Awards & Community


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:

  • Computational geophysics
  • Inverse problems
  • Seismic imaging and parameter inversion
  • Seismic processing
  • Machine learning
  • High-performance computing
  • Scientific software development
Here are some of the research topics that you will be able to contribute to:

  • Geophysical target-oriented imaging and model parameter inversion
  • Applications of deep learning to physics-driven inverse problems
  • Large-scale inverse problems using HPC hardware
If you find any of these topics exciting and would like to participate in the development of these new theories and algorithms, send me an email!

By joining our group, you will be part of a vibrant community of computational and experimental geoscientists in the Physical Science and Engineering division. As part of your Ph.D. programme you will get the opportunity to take classes from world-renowned professors, and you will have access to Shaheen II, one of the most powerful supercomputers in the world.

Through my wide network of academic and industry connections, you will also be able to spend some of your time in other institutions and/or take on summer interships in other parts of the world. Finally, you will learn a great deal of scientific software development and contribute to the development of open-source tools such PyLops, and create new ones. Be prepared to use Python, C/C++, and CUDA - but do not be discouraged to apply if you are not familar with them, yet!

More details about our graduate program can be found here.