Nathaniel Le Sage Email & Phone Number

United States

Made over 50 contributions to Wikipedia articles since 2011.https://en.wikipedia.org/wiki/User:Krunchyman

Austin, Texas, United States

Back-end and machine learning engineer for legal tech startup Novo. Worked with Next.js, Node, Redis, PostgreSQL, Supabase and Puppeteer. Wrote back-end PostgreSQL queries and file streaming logic. Also wrote scripts for web scraping and converting LLMs to various file formats (e.g. HF to GGUF), and a Python HTTP server for LLM embeddings. Worked in lean, fast-paced environment with small team, low overhead and quickly changing product requirements.

Wrote onboard licensing software and extensive unit test suite reaching >90% coverage, for embedded displays deployed across 15 product models. Developed expertise in C++, CI/CD, Qt framework, Google Test and CAN bus protocol.

Bettendorf, Iowa, United States

Used decision trees and autoencoders predict stock tail risk, training them on price and fundamental data. Although the datasets were on different timescales, I used pandas to resample and align them. I used fuzzy matching to match the datasets by their stock name columns and employed a variety of descriptive statistics and evaluation metrics, such as accuracy, precision, recall and F1 score, to analyze the data and evaluate the models.Finally, I used memoization and joblib CPU… Show more Used decision trees and autoencoders predict stock tail risk, training them on price and fundamental data. Although the datasets were on different timescales, I used pandas to resample and align them. I used fuzzy matching to match the datasets by their stock name columns and employed a variety of descriptive statistics and evaluation metrics, such as accuracy, precision, recall and F1 score, to analyze the data and evaluate the models.Finally, I used memoization and joblib CPU parallelization to speed up intensive data processing tasks with 4x and 12x improvements in speed, respectively. For the company's infrastructure, I wrote a class to handle a series of tasks involving converting .sqlite files into .csvs and moving them from their original nested directory structure into another structure that matched that of our previous data product. This let us reuse code written for that product. Show less

Events Chair officer at ACM student chapter at the University of Iowa. Organized UICC 2020.

Arroma

Student research assistant responsible for various assignments at the University of Iowa ARROMA lab.In the summer of 2018, I worked on an ML model that predicted PM2.5 concentrations based on data from NASA's VIIRS DNB and MERRA-2. Our model used random forests and gave a cross-validation MAPE of 0.18.In the spring 2018 semester, I created and documented an atmospheric sciences software library with Sphinx. Some of the features of this library included functions to calculate and… Show more Student research assistant responsible for various assignments at the University of Iowa ARROMA lab.In the summer of 2018, I worked on an ML model that predicted PM2.5 concentrations based on data from NASA's VIIRS DNB and MERRA-2. Our model used random forests and gave a cross-validation MAPE of 0.18.In the spring 2018 semester, I created and documented an atmospheric sciences software library with Sphinx. Some of the features of this library included functions to calculate and generate plots of solar irradiance, Pascal's law, and Wien's displacement law, among other things. I also worked with the AirNow API and the CheckWX API, which provide weather forecasting data, and METAR and TAF reports. I wrote several Python scripts that retrieve forecasts from the AirNow API.I regularly used Python, NumPy, matplotlib, OpenCV, scikit-learn, TensorFlow, Keras, Linux shell and HPC software to accomplish geoscience-related tasks. Show less

Co-op programmer at Bio::Neos. Worked on two lab information management systems in Drupal and Ruby on Rails, putting them both into production. Made over 100 Git contributions. Used Git, HTML, JavaScript, Drupal, Ruby on Rails and Ember.js. Created UIs, wrote text processing algorithms and made SQL queries for pulling information from databases. Also wrote documentation and created process flowcharts.In my first week, I worked with another student on a bioinformatics application in Java… Show more Co-op programmer at Bio::Neos. Worked on two lab information management systems in Drupal and Ruby on Rails, putting them both into production. Made over 100 Git contributions. Used Git, HTML, JavaScript, Drupal, Ruby on Rails and Ember.js. Created UIs, wrote text processing algorithms and made SQL queries for pulling information from databases. Also wrote documentation and created process flowcharts.In my first week, I worked with another student on a bioinformatics application in Java and JavaFX. The app's functionality included methods for uploading .fasta files, converting DNA representations into mRNA and protein representations, and sequence alignment. To do the sequence alignment, I implemented the Needleman-Wunsch global alignment algorithm. Show less

Student programmer who worked on two healthcare enterprise applications in the .NET framework and put them into production. I used such technologies as C#, WPF, XAML, and NUnit. I also maintained a Microsoft SQL database belonging to the healthcare system. I gained experience in good software development principles and employed object-oriented design patterns such as abstract factory and prototype.

Bachelor Of Arts - Ba, Computer Science

Minor, Mathematics