MARY SILVA
×

About Me

2019 LLNL Poster Symposium

Mary Silva

I recently obtained my Masters in Statistics and Applied Math at University of California, Santa Cruz. I received my B.S. in Applied Math with emphasis in computer science from San Francisco State University. My interests are deep learning and statistical modeling for climate and enviromental science related data. I am currently an intern in the Data Science Summer Institute at Lawrence Livermore National Labs.

My masters capstone project involved building Gaussian process regression models to learn about the effects of climate on global vegetation. I was advised by Raquel Prado and Kai Zhu.

mail: msilva8@ucsc.edu


Resume

Lawrence Livermore National Labs - DSSI (June 2019-Present)
  • Analyzed and processed large NetCDF climate data files from various agencies, which spanned 1979-2017 at 6-hour intervals with a resolution of 0.5×0.5.
  • Managed and transferred climate data from Analytics and Informatics Management Systems GPU to the Livermore Computing institutional Bitbucket.
  • Developed Bayesian hierarchical models for hurricane data using R in order to detect changes in hurricane behaviors; specifically, to determine if the overall movement of hurricanes has changed over time and what underlying factors contributed to these changes.
  • Presented and submitted a poster for the Livermore 2019 Poster Symposium describing hurricane analysis, methodologies, and results.
  • Authored reports for senior machine learning and climate scientists from the National Atmospheric Release Advisory Center.
Lawrence Livermore National Labs - Computation Scholar (June 2018-October 2018)
  • Implemented discrete- and continuous-time hidden Markov Models using Python for analysis of medical data.
  • Evaluated recently developed Python probabilistic programming languages such as PyMC3 and Edwards, which utilize Theano, Tensorflow and PyTorch, for efficiency comparison of complex statistical models.
  • Developed Bayesian hierarchical models for hurricane data using R in order to detect changes in hurricane behaviors; specifically, to determine if the overall movement of hurricanes has changed over time and what underlying factors contributed to these changes.
  • PPresented and discussed the benefits and weaknesses of using these high level probabilistic programming languages for senior machine learning staff.
Modesto Irrigation District, Finance Analyst Intern (2016)
  • Used customer service and generation facilities data to develop pricing strategies
  • Developed regression and classification models for prediction and inference of energy consumption
  • Analyzed bond documents and financial statements spanning several decades
Del Monte Foods (2012-2015)
  • Checked warehouse inventory for proper location storage and proper labeling of product
  • Verified ERP reports, BPCS, and inventory databases were up to date
  • Analyze hourly production and inventory transactions

LinkedIn

2019 LLNL Poster Symposium - Hurricane Analysis Project.