Portfolio

Several significant achievements across technology, banking, retail lending, and market research sectors are outlined below. The blend of technical skills, strategic thinking, and innovative problem-solving I've demonstrated illustrates my leadership and expertise in the field. Further down this portfolio highlights relevant projects of various sizes, skills, and certificates.

Achievements

  • Successfully submitted to the Federal Reserve a set of time-series and customer level predictive stress testing models on a $14B mortgage portfolio, which support capital planning for Fifth Third Bank (CCAR / DFAST).
  • Drove over $20MM of incremental value in loss reduction through customer level predictive modeling of origination and fraud scorecards. Models supported the underwriting of nearly $1 billion in loans.
  • Generated $1M+ in revenue by developing an automated predictive modeling engine that won P&G's global forecasting business for The Nielsen Company. Platform deployed to Nielsen hubs globally, also winning non-P&G business in Western Europe.
  • Nielsen also used the forecasting platform to secure their role as Walmart’s primary insights partner.
  • 4 US patent applications on an automated business alerting platform, one of Nielsen’s top ten initiatives of 2012.
  • Conducted well over 100 interviews across multiple organizations to hire and manage high-performing data science teams that align strategic vision with technical resources required to achieve business objectives.

Projects

Origination Model (Credit Risk)

This project applies various classification models such as Logistic Regression (pytorch & sklearn), Random Forest, and XGBoost to solve an import origination problem in the credit risk space. The models for the most part have comparable accuracy metrics aside from xgboost, which may perform better with some additional fine tuning.

Basket Analysis (Retail Analytics)

This Instacart basket analysis is focused on understanding customer purchasing patterns using a dataset containing 3 million grocery orders from over 200,000 Instacart users. The files encompasse a variety of data on order details, product information, and the order in which products were added to the cart.

Slack Alerting for AWS

It's really easy for AWS costs to slip under the radar and accrue to unexpected amounts. While Amazon does provide some functionality with their cost explorer meant to address that, it is not as flexible in how the data can be aggregated and it isn't integrated with Slack as far as I'm aware. This project addresses that.


Micro Projects

Core Competencies

  • Languages: Python, SAS, SQL, R
  • Methodologies: Machine Learning, Time Series Analysis, Deep Learning, Statistics, Explainable AI
  • Tools: Amazon Web Services (S3, Lambda, ECR, ECS, EC2, Cloudwatch), Docker, Airflow, Elasticsearch, PostgreSQL, Flask, MS Excel, Tableau

Certificates