Data Scientist

• Aug 2019 - Present

U.S. Soccer Hackathon Finalist

• Jun 2018 - Jun 2018

Created finalist project in Official U.S. Soccer Hackathon involving the Implementation of k-means clustering of players into squad roles using Opta positional data in order to Demonstrate differences in roster spot value across positional groups Created finalist project in Official U.S. Soccer Hackathon involving the Implementation of k-means clustering of players into squad roles using Opta positional data in order to Demonstrate differences in roster spot value across positional groups

Data Scientist

• Apr 2017 - Jul 2017

Metis is Data Science bootcamp that teaches machine learning, statistics, programming, data visualization, and design through a project based curriculum. In three months, I completed five data science projects from data acquisition to presentation including data cleaning, exploratory analysis, supervised and unsupervised machine learning, natural language processing, and web app development. Projects were as follows: 1.) Developed a node.js app to generate real time predictions of… Show more Metis is Data Science bootcamp that teaches machine learning, statistics, programming, data visualization, and design through a project based curriculum. In three months, I completed five data science projects from data acquisition to presentation including data cleaning, exploratory analysis, supervised and unsupervised machine learning, natural language processing, and web app development. Projects were as follows: 1.) Developed a node.js app to generate real time predictions of opponent’s future cards in Hearthstone using a 2 layer « card2vec » neural network trained on previous card sequences attained from the Track-o-Bot API 2.) Produced an NLP topic model of 6 million tweets from popular political journalists and bloggers using latent dirichlet allocation and latent semantic indexing, applied word2vec sentiment analysis to each user, and clustered users by sentiment with a k-means algorithm 3.) Analyzed geographic and water quality features of water sources in Tanzania to classify functional status using a random forest classifier implementing a Naive Bayes model to impute missing feature values to correctly classify the functional status of 80% of sources 4.) Developed a codependent expected goal difference model using linear regression of match features from WhoScored.com in order to demonstrate distinction in feature significance between home and opposition 5.) Analyzed MTA turnstile data to optimize deployment of canvasing street teams in NYC Show less Metis is Data Science bootcamp that teaches machine learning, statistics, programming, data visualization, and design through a project based curriculum. In three months, I completed five data science projects from data acquisition to presentation including data cleaning, exploratory analysis, supervised and unsupervised machine learning, natural language processing, and web app development. Projects were as follows: 1.) Developed a node.js app to generate real time predictions of… Show more Metis is Data Science bootcamp that teaches machine learning, statistics, programming, data visualization, and design through a project based curriculum. In three months, I completed five data science projects from data acquisition to presentation including data cleaning, exploratory analysis, supervised and unsupervised machine learning, natural language processing, and web app development. Projects were as follows: 1.) Developed a node.js app to generate real time predictions of opponent’s future cards in Hearthstone using a 2 layer « card2vec » neural network trained on previous card sequences attained from the Track-o-Bot API 2.) Produced an NLP topic model of 6 million tweets from popular political journalists and bloggers using latent dirichlet allocation and latent semantic indexing, applied word2vec sentiment analysis to each user, and clustered users by sentiment with a k-means algorithm 3.) Analyzed geographic and water quality features of water sources in Tanzania to classify functional status using a random forest classifier implementing a Naive Bayes model to impute missing feature values to correctly classify the functional status of 80% of sources 4.) Developed a codependent expected goal difference model using linear regression of match features from WhoScored.com in order to demonstrate distinction in feature significance between home and opposition 5.) Analyzed MTA turnstile data to optimize deployment of canvasing street teams in NYC Show less