$8.5M+ in value Built an occupancy predictor ML model for the City of Detroit which, among many other things, allowed the city to focus efforts on helping Detroit’s buy back foreclosed homes, valued at over $8.5M In public benefits

An analytics / data science / data engineering Swiss army knife at a fast paced startup - I interface with clients, providing both insights and training to help them make better data driven decisions protecting infrastructure, as well as continuing to build, deploy, and monitor our ML risk models. I've also used this client interaction to work cross functionally and give training and feedback to internal business partners (Customer Success, Machine Learning, App development teams, Sales, Executive leadership team). Additionally, I continue to build out our end to end ETL pipelines to ensure up-to-date, high quality data that make both the model builds and analysis quick to produce and implement. I do all of this work using a combination of python, Google's suite of tools (GCS, Google Data Studio, Trifacta, Alteryx, BigQuery, Looker), Tableau, and AWS.

Writing production code / building ML models with a focus on energy utilities and their gas assets - extremely collaborative (agile) process, saving lives and utility companies millions of dollars due to significant reduction in number of incidents / gas explosions

Performed GIS R&D and built lightweight POC products. Developed Urbint's first spatial impact model under immense pressure to our biggest customer, ensuring that they were able to protect their most vulnerable communities and assets as COVID intensified

AV / Smart Cities research - Provided data science expertise to support Ford Smart Mobility products and programs, such as an accessibility tool to support AV initiatives, as well as year long engagements with depts. of transport in Mexico City, Miami, Pittsburgh, and Detroit.

-Improved the granularity of predicting household waste generation for the Department of Sanitation New York (DSNY) by building a neural network model with an R-squared nearing 0.87.-Improved NYPD’s ability to predict the propensity for a neighborhood to report a shooting incident by building a random forest classification model.

Capstone project reduces municipalities’ departments of transportation costs by 95% in assessing bike lane quality, while time savings run into the thousands of hours. Accomplished through computer vision algorithms, crowd sourced data collection, and cloud computing.

Saved LS&Co. $5 million thru time series forecasting models to measure stock-outs - published dashboards and presented results to the President of Levi’s, planning teams, and merchandising teams.

Reduced the time taken in bringing cases to court by DOJ lawyers by 50% by building a logit classifier model detecting providers committing Medicare fraud.