Software Developer
JONATHAN DONESKY
Phone: 301-***-**** Email: adkfmq@r.postjobfree.com
LinkedIn: /jonathan-donesky-539000127
Github: /jdonesky
Technical Skills
Front end
• React/Next.js, Redux,
ES6/JavaScript, Typescript
• HTML5, CSS, Materialize, EJS
Back end
• Python, Node, Express,
GraphQL (Apollo)
Databases
• MySQL, Firebase, MongoDB,
AWS Elastic Beanstalk, S3
DevOps
• Git, Github, Docker,
Kubernetes, Jenkins
Testing
• Jest, Enzyme
Data Science & Machine Learning
• Keras, Tensorflow, Pytorch
Software Projects
JSCRATCH - Software used: Typescript, React (hooks), Redux, Express.js, esbuild, Wasm, Webpack, npm January 2020 In-browser coding environment with real-time bundling and transpiling of ES6, JSX and CSS
• Achieved lightning-fast bundling and transpiling in-browser by customizing plugins to esbuild and Wasm
• Integrated advanced IDE features such as syntax highlighting, autocomplete, and code linting using Monaco Editor.
• Integrated custom redux middleware containing debouncing logic to optimize performance on save-data actions
• Reduced latency through caching queried modules in browser’s indexedDB.
• Added security layer isolating execution of user-provided code in iframes and restricting communication with the parent.
• Streamlined dependency management using multi-package architecture with symlinks. DUMB FACEBOOK - Software used: React (16.8 - Hooks), Redux, React-router, Firebase, Webpack, npm December 2020 Single page web app mimicking styles and functionality of Facebook (link)
• Combined Firebase Realtime database and Cloud functions to create reactive, real-time data storage.
• Designed and implemented persistent data layer using Redux and localStorage.
• Reduced page load latency through code-splitting using higher order components and dynamic imports.
• Integrated APIs such as Google Maps, providing location search autocomplete and geocoding with a responsive map interface.
• Leveraged Github Actions in CI/CD pipeline.
FLASH - Software used: Python, Keras, Tensorflow, Pandas, numpy, scikit-learn, matplotlib Jan 2020 - April 2020 Long Short-Term Memory Recurrent Neural Network, models flash-flooding in Ellicott City, MD
• Used LSTM layers to capture time-lagged flood factors and reduce backpropagation of errors.
• Reduced noise by interpolating for missing and false stream sensor readings in training data.
• Prevented overfitting by incorporating dropout regularization (algorithm probabilistically excluding weight updates). Work Experience
SOFTWARE ENGINEER INTERN • NASA Goddard Space Flight Center • Greenbelt, Maryland Jan 2020 - April 2020
● Developed a Long-Short Term Memory (LSTM) neural network using Keras and Tensorflow to forecast flash floods and improve evacuation warning systems in Ellicott City, MD.
● Increased forecast accuracy by 14% (Nash-Sutcliffe Efficiency score) using model sensitivity analysis to determine optimal combination of environmental inputs.
● Created data pipeline with Python/Pandas to integrate, process, and visualize 10 years of satellite and in situ sensor readings. PROJECT MANAGER • Laisar Management Group • Silver Spring, Maryland Nov 2018 - April 2019 / Nov 2019 - Feb 2020
● Managed technology research, supply chain analysis, and regulatory compliance for Offshore Wind developer Ørsted.
● Applied domain knowledge to develop software for tracking employee demographics, work duties, and hours. Education
B.S. ENVIRONMENTAL SCIENCE AND POLICY • University of Maryland • College Park 2017 – 2020
● Honors: magna cum laude (GPA 3.96/4.0)