Unitarian Universalist Legislative Ministry of New Jersey · Congregational Liason |
2015 to Current
|
Data scientist with skills in exploratory data analysis, graphical analysis, statistics, hypothesis testing and predictive analytics. Flexible and eager to learn new domains and skills. Generates reliable solutions by combining industry insights with analytics and verification of underlying assumptions.
Metis |
New York, NY
|
Data Scientist |
Sep 2016 to Dec 2016
|
- Completed 5 projects covering the entire data science process including inquiry formulation, data acquisition and munging, model fitting, evaluating, and interpretation, data visualization, and presentation
- Tools used include python, jupyter notebook, scikit-learn, statsmodels, pandas, matplotlib, seaborn, beautifulsoup, AWS, MongoDB, git and github
Bridg-it LLC |
New York, NY
|
Data Scientist |
2015 to 2016
|
Full-time consultant responsible for predictive analytics strategy, data architecture and hands-on implementation for this Ed-Tech start-up.
- Developed data integration strategy/database architecture for predictive analytics, data warehouse, and dashboards
- Built logical schema for data warehouse (star schemas) and worked with DBA on physical implementation (MySQL, Orient db)
- Created dashboard data visualizations (R, AMCharts and D3.js) including design, SQL, and JavaScript package implementation
- Deployed nginx, Shiny, D3.js apps in Windows and linux (Centos) environments
Credit Suisse |
New York, NY
|
Business Analyst |
2004 to 2014
|
- Provided SQL solutions for complex Control M generated Oracle relational database (RDBMS)
- Optimized application strategy using analysis of user behavior logs and operations data using trends correlation analysis, ROI, information visualization, business analytics, etc.; Data visualization with JavaScript-based VML
G.E. Capital (Deutsche Financial Services) |
New York, NY
|
Systems Analyst |
2000 to 2003
|
Universal Instruments |
Binghamton, NY
|
Statistician |
1995 to 2000
|
Predicting Poverty Extent Via Community Factors |
|
Used classification machine learning methods (logistic regression, naive bayes, and random forest) to predict high or low levels of poverty in US counties. Created interactive data visualization of county-level US poverty rates. http://mitchki.com/D3/poverty.html
Text Mining Analysis: Data-Focused Job Listings |
|
Used Latent Dirichlet Allocation (LDA) methodologies to identify latent topics in 27,000 job listings scraped from the indeed.com job board. Naive Bayes and Random Forest algorithms were used to predict specific keywords and job locations, and verified via precision/recall metrics and confusion matrix.
Data Seedlings in the Garden State |
|
K Means and hierarchic unsupervised algorithms used to identify New Jersey municipal economic clusters; evaluated via silhouette score and graphical analysis.
Assess Demand for Your Skills in Major US Cities |
2015
|
Real-time polling of Indeed job board to quantify demand for any skill in major US cities.