Skip navigation

Welcome to my personal homepage

Subscribe to News RSS feed

Welcome to my personal website. I work on various projects for prescriptive data analytics utilizing machine learning (ML) and deep learning (DL). Before coming to industry, I spent several years at Stanford University, studying statistical learning and optimization and doing research in massive computational experiments for prescriptive analytics. My dissertation includes advances in closed-loop optimization, application of unsupervised learning for accelerating stochastic programming (massive computational experiments), and prescriptive analytics framework based on multiple economic measures (e.g., both the NPV and rate or return).

I started my undergraduate study in 2004 at Sharif University in mechanical engineering. I was very curious about subsurface resources. It was amazing how the oil industry has shaped the modern world... if there was no oil industry, the auto industry would not have flourished. Perhaps we did not have airplanes either, and there was no computer or internet! How did our earth create and maintain the oil for millions of years? and how is the human able to drill 5-mile holes and extract the black-gold? I started studying chemical & petroleum engineering as a second major in 2005, driven to answer these questions. First thing I had realized, the oil is maintained in some porous rock, and not in some tank-shaped empty space!

In 2009, I went to Tulsa to study for MS degree at University of Tulsa (TU). There I pursued my passion towards numerical optimization: how do we formulate a decision making process in terms of mathematical objectives and constraints and develop efficient computational algorithms to solve a massive computational experiments. My particular focus at TU was developing efficient TSVD-based nonlinear optimization algorithms for solving massive nonlinear inverse problems. The results on that effort was published in two journal papers on TSVD-based optimization for solving massive nonlinear inverse problems. During MS study, I also worked on other interesting topics in such as derivative-free and gradient-based large-scale optimization and ensemble-based data assimilation methods.

Visualization Codes for ADGPRS simulations

I started uploading some visualization/post-processing codes which I wrote in the past few years.

If you are using ADGPRS for your simulation, you may find some useful codes here.