MAkE iT HApPeN! 

Webpages:

LLNL: https://people.llnl.gov/fernandez48

LinkedIn: www.linkedin.com/in/mariagisellefernandez

Google Scholar: https://scholar.google.com/gisellefernandez

Work email address: fernandez48@llnl.gov

Alternative email address: mariagisellefernandez@gmail.com

About me  -- Math decorates your mind La matemática te adorna la cabeza -- Giselle

My name is Giselle Fernández and I am scientist. I do math, physics, and modeling. I extract hidden patterns from the real world and simulations data. I have a unique set of skills, including nuclear physics, machine learning, uncertainty quantification, and data analysis. I can make efficient models and report confidence bounds. I am a team player, keep the ball rolling, and am enthusiastic and eager to learn.

More about me ...

I am currently a Data Scientist working for the Atmospheric Science Research & Applications Group (AEED, PLS) at Lawrence Livermore National Laboratory.

I was a postdoctoral researcher at Los Alamos National Laboratory (LANL). One of the main missions of LANL is to protect the nation by ensuring the safety and reliability of weapons. Understanding material properties and behavior is a key component of this mission, for example, it is critical to characterize brittle metal behavior and understand how failure occurs. Being able to predict crack propagation and interaction allows for potential accidents to be prevented. Although in some cases, experiments can be done, we cannot afford more than a few. Multiphysics multiscale highly resolved models might be also available, but the computational cost associated makes them prohibitively expensive for uncertainty quantification, where millions of simulations are needed. I am working on constructing multi-fidelity surrogates aided by machine learning to quantify uncertainty in fracture models. If these surrogates are constructed efficiently, they can reduce the computational cost by orders of magnitude while maintaining the desired accuracy. Consequently, uncertainty quantification, optimization, and inference can be achieved.

I received my Bachelor of Science in Nuclear Engineering from Balseiro Institute, San Carlos de Bariloche, Argentina in 2014. My thesis title is "Development of a surveillance program for the Argentinian reactor CAREM 25". I received my Master of Science in Mechanical Engineering in 2016 and my doctoral degree in Aerospace Engineering in 2018 at the University of Florida in Gainesville, Florida.  My thesis title is "Quantifying particle departure from axisymmetry in multiphase cylindrical detonation".

I was a member of the Center for Compressible Multiphase Turbulence in the Structural and Multidisciplinary Optimization group for the PSAAP II project. At the CCMT center, we worked as a team to predict multiphase detonations using large-scale simulations. My work was focused on uncertainty quantification for macroscale multiphase simulations. I used high-performance computers to run simulations in parallel. I worked with programming languages such as Matlab, Python, Fortran, C++, and Bash to facilitate the submission and processing of the data. 

I have a broad expertise background: 2 years of Mathematics degree, 3 years of Physics degree. A B.S. degree in Nuclear Engineering. An M.S. degree in Mechanical Engineering and a Ph.D. degree in Aerospace Engineering. My Ph.D. thesis was at the intersection of flow physics, very large-scale simulations, surrogate models, optimization, and uncertainty quantification. This is reflected in the fact that one of my advisors, Dr. Balachandar, has expertise in computational multiphase flow, direct and large eddy simulations of transitional, turbulent flows, and integrated multiphysics simulations of complex problems, while the other, Dr. Haftka, has expertise in optimization methodology applied in structural design including sensitivity calculation and approximation techniques.

Interests:  "Anomaly Detection via Groups of Simulations", "Hidden Noise in Simulations", "Uncertainty Quantification", "Optimization", "Machine Learning", "Multi-fidelity Surrogate Models"