William Grisaitis, Machine Learning Developer in Orlando, FL, United States
William Grisaitis

Machine Learning Developer in Orlando, FL, United States

Member since July 30, 2018
William is a data scientist with a diverse background and skill set, having worked on deep learning, time series forecasting, Bayesian modeling, and more. He has co-authored academic research applying deep learning to computational neuroscience and has worked in industry applying machine learning to financial markets and consumer finance. William graduated from Duke University where he studied physics, finance, and literature.
William is now available for hire

Portfolio

Experience

  • Python, 10 years
  • NumPy, 8 years
  • Pandas, 6 years
  • Scikit-learn, 6 years
  • Jupyter Notebook, 4 years
  • Docker, 3 years
  • Caffe, 3 years

Location

Orlando, FL, United States

Availability

Part-time

Preferred Environment

Mac or Linux, Python IDEs, Jupyter Notebooks

The most amazing...

...project I've worked on was mapping neurons in the fruit fly brain with convolutional neural networks.

Employment

  • Data Scientist/Quantitative Analyst

    2018 - PRESENT
    A Hedge Fund in the United States
    • Developed predictive models of electricity prices in the United States. Developed trading strategies.
    • Conducted exploratory analysis to understand market behavior and predictive variables.
    • Incorporated external data sources to improve prediction accuracy, trading profitability.
    • Developed a pipeline to streamline data ingestion.
    • Built new predictive models on an ongoing basis.
    • Analyzed time series with correlation analysis, k-means clustering, and other methods.
    Technologies: Python data science libraries, AWS, Jupyter
  • Senior Software Engineer

    2016 - 2017
    Howard Hughes Medical Institute Janelia Research Campus
    • Trained convolutional neural networks that learn to map neurons in images of brain tissue.
    • Developed complete data science pipeline, including data storage, support code for training, cluster deployment, and evaluating results (metrics, plots).
    • Co-authored publications about work.
    Technologies: Caffe, Python data science libraries, Docker, Nvidia GPUs, Linux
  • Senior Associate

    2013 - 2016
    Capital One
    • Conducted financial and statistical analysis for business decision making in consumer/small business finance.
    • Developed software for an internal data science application, using React and Python on AWS.
    • Filed a patent for a self-driving car with a built-in ATM.
    Technologies: Python, SQL, MS Office, JavaScript

Experience

  • Used Deep Learning to Infer Brain Structure in Fruit Fly Brain Imagery (Development)
    https://ieeexplore.ieee.org/document/8364622

    I was a lead engineer and co-author on published research applying deep learning to neuroscience. With other researchers, I developed a deep learning approach to "map" neurons in images of fruit fly brain tissue.

    My responsibilities include all aspects of the project:
    * deep learning - training and evaluating neural networks. Developing support code for automating and repeating this work.
    * data engineering - storing large image data, developing access layers
    * sys admin - maintaining a cluster of GPU machines. Deploying code with Docker and Mesos. General Linux/Ubuntu sysadmin.
    * miscellaneous - data visualization, Jupyter, visualizing results, etc.

  • Forecast Electricity Prices in Wholesale Energy Markets for Financial Trading (Development)

    In this work, I produced predictive regressions of energy prices in wholesale markets, where electricity is bought and sold in a futures market. My results are currently (Oct 2018) being used for financial trading by a hedge fund in North Carolina, USA.

    I worked with domain experts to develop a predictive linear model that successfully explains about 20% of all price movement in the market. Beyond statistical modeling and analysis, I also prepared all data needed for the project, starting from the ground up. This included data about the market itself as well as other data relevant to the problem, such as weather data.

Skills

  • Languages

    Python, Bash, SQL, R
  • Frameworks

    Caffe
  • Libraries/APIs

    Scikit-learn, Pandas, NumPy, HDF5, Matplotlib, TensorFlow
  • Platforms

    Jupyter Notebook, Docker
  • Other

    Machine Learning, Optimization, Linear Algebra, Bayesian Statistics

Education

  • Bachelor of Arts degree in Physics (major), Economics and Literature (minors)
    2007 - 2012
    Duke University - Durham, North Carolina

To view more profiles

Join Toptal
I really like this profile
Share it with others