Bio


I was born and brought up in India, where I completed my undergraduate education. Upon graduation, I joined Schlumberger and worked on high-end imaging applications in USA and Mexico. After spending 4 years in the industry, I decided to return to graduate school and completed a Masters in Computational and Mathematical Engineering at Stanford university in 2017. I'm currently a Ph.D graduate student in the same department, with my primary interests in convex and non-convex optimization.

Honors & Awards


  • Shell Fellowship, Stanford University (2015-2016)
  • DAAD WISE Scholarship, DAAD (Deutscher Akademischer Austauschdienst) (2010)
  • Institute Silver Medal, Indian Institute of Technology, Kharagpur (2011)

Education & Certifications


  • Ph.D., Stanford University, Computational and Mathematical Engineering
  • MS, Stanford University, Computational and Mathematical Engineering (2017)
  • Integrated BS and MS, Indian Institute of Technology, Kharagpur, Geophysics (Major), Physics (Minor) (2011)

Stanford Advisors


Personal Interests


Traveling, Physics, Finance.

Current Research and Scholarly Interests


My current research interests are related to a category of large scale, convex and non-convex optimization problems that arise in the context of waveform inversion.The non-convexity of the problem presents interesting mathematical challenges, while the huge amounts of data that need to be processed present significant computational challenges.

I'm currently looking at frequency domain methods using sparse matrix factorization algorithms to solve the Helmholtz equation, in order to computationally speed up the solution to these large scale optimization problems, which often involve a million to a billion variables.

I'm also looking at several Artificial Intelligence and Machine Learning techniques which could help solve some interesting problems in Geophysics, where such techniques are yet to create any significant impact. In particular, I'm looking at using techniques from topological data analysis as a tool for detecting qualitative features in large scale seismic data sets.

Projects


  • Finding a cover for an ellipse with N rectangles, Stanford University (1/1/2016 - 3/31/2016)

    This is an interesting class project that I did as part of my "Numerical Optimization" class at Stanford. The problem statement is to find a cover for an ellipse with N rectangles, such that the area outside the ellipse is minimized. In this project, I first formulate an equivalent problem that reduces to finding a cover for a circle with the same number of rectangles. The problem is then solved using a Modified-Newton Hessian based approach, and the performance is compared against Steepest Descent. It is found that Modified Newton based approach is much more efficient, although each iteration is significantly more expensive compared to Steepest Descent. I also compare the performance of using different line search algorithms like Goldstein vs Strong-Wolfe conditions, and conclude that the Strong-Wolfe conditions provide much better results.

    Location

    ICME, Stanford University

    For More Information:

  • Dynamic Asset Allocation using Reinforcement Learning, Stanford University (9/20/2016 - 12/16/2016)

    This was a project done as part of the CS221 (Artificial Intelligence) class at Stanford University. In this project, we used reinforcement learning to perform asset allocation between two asset classes - a bond and a stock. Model based and model free techniques were used to perform this task, and the results were compared against one another and also against historical returns of the stock.

    Location

    ICME, Stanford University

    Collaborators

    • Enguerrand Horel, Ph.D. Student in Computational and Mathematical Engineering, admitted Autumn 2016, School of Engineering
    • Victor Storchan, Software Engineer, Adobe

    For More Information:

  • Automated Aircraft Touchdown, Stanford University (10/1/2016 - 12/31/2016)

    This was a class project done as part of the "Decision Making Under Uncertainty" class at Stanford University. In this project we experiment with a few Reinforcement Learning algorithms with the goal to safely land an aircraft, in the presence of stochastic winds. The project was implemented in python 2.7. Code on github: https://github.com/rsarkar-github/CS238-Automated-Aircraft-Landing-Reinforcement-Learning

    Location

    ICME, Stanford University

    Collaborators

    • Amy Shoemaker, School of Engineering
    • Sagar Vare, School of Engineering

    For More Information:

  • Information Directed Reinforcement Learning, Stanford University (1/1/2017 - 3/30/2017)

    This is a project done as part of CS334 (Advanced Reinforcement Learning) class at Stanford University. In this project, we explored an efficient exploration strategy based on information directed reinforcement learning. Details are provided in the attached paper.

    Location

    ICME, Stanford University

    For More Information:

Work Experience


  • Quantum Algorithms Researcher, QC Ware Corp. (7/1/2018 - 9/23/2018)

    Research into quantum algorithms on finance, and topological data analysis.

    Location

    Palo Alto, USA

  • Application Developer, QC Ware (7/1/2017 - 9/25/2017)

    Worked on binary optimization problems that can be solved using a quantum computer. Current projects include : Financial Applications, Topological Data Analysis.

    Location

    Palo Alto

  • Geophysicist, Schlumberger (10/1/2013 - 8/31/2015)

    My role in this position is to foster growth of the new and advanced technology businesses in Mexico, which include Full Waveform Inversion, Seismic Guided Drilling, Advanced Depth Imaging among others. Other responsibilities include training the staff in using these technologies, assist the center in winning new businesses and help in the execution of these projects.

    Location

    Mexico

  • Incubator Program, Schlumberger (7/24/2011 - 9/30/2013)

    I was part of a cohort of ~10 people recruited globally for a special incubator training program. Worked with advanced seismic imaging algorithms and model parameter estimation using full waveform inversion and ray based tomography.

    Location

    Houston, USA

All Publications


  • Snell tomography for net-to-gross estimation using quantum annealing SEG 88th Annual Meeting Sarkar, R., Levin, S. : 5078–82