Sep 2017 to Dec 2017
Metis is a full-time, ACCET accredited 12 week data science program that focuses on Python, machine learning, statistical modeling, big data tools, and building a professional portfolio of five projects using these tools and techniques. Selected Projects:
Mix Retriever: A Hip-Hop Playlist Generator
Tools and Techniques: Doc2vec, SVD, Sentiment Analysis, Nearest Neighbors, AWS.
- A content focused playlist generator that builds a hip-hop playlist of songs with similar lyrical meaning and mood to a song specified by the user.
- Trained a Doc2vec model on AWS from a corpus of 30,000 songs.
- Data collected from Wikipedia, Spotify API, and LyricWiki.
Would Reddit Like My Comment?
Tools and Techniques: MongoDB, Lemmatization, TFIDF, Sentiment Analysis, Flask.
- Collected 200,000 Reddit comments and their scores from multiple subreddits with the goal of predicting what type of comments would score highly on those subreddits.
- Used a Decision Tree algorithm with recall scores between 60% and 85% depending on subreddit.
- Data collected using PRAW (Python Reddit API Wrapper).
Supervised Learning and Online Gaming
Tools and Techniques: classification algorithms (SVM, Decision Tree, Random Forest, Logistic Regression, Bernouilli Naive Bayes), PostgreSQL.
- Analyzed various classification algorithms to build model predicting the winner of a match from popular online video game, League of Legends.
- Final model used SVM algorithm and had an accuracy of 71.3%.
- Data downloaded from Kaggle.
|Software Developer (Intern)||
Jan 2017 to May 2017
Real Time Ad Campaign Management
Analyzed automotive marketing data to improve marketing campaign effectiveness by making specific recommendations, such as removing ineffective customers from mailing lists and predicting which customers would be most likely to purchase a vehicle. Specific contributions included:
- Co-lead student team and corporate clients throughout duration of the project.
- Designed the graph model of exposure, lead, and sales data for multiple marketing campaigns within a Neo4j graph database.
- Wrote Python scripts that accurately mock and update marketing campaigns’ data by simulating realistic and random real time changes for these campaigns.
- Created data models on C# backend utilized by frontend, and deployed using Docker.
|Java Software Developer (Intern)||
Jun 2016 to Aug 2016
Personal Property Team:
- Updated insurance policy cancellation webpages fulfilling business requirements from multiple departments to make the cancellation process smoother for users.
- Re-wrote server logic involving handling error pages for in-house software. Changes let employees know what error was being reported more clearly.
- All software testing done with JUnit.