Jeffrey G. Email & Phone Number

United States

San Francisco, Ca, Us

Mclean, Va, Us

Unleashing the potential of Azure ml studio & Azure Databricks to provide enterprise level DataOps & MLOp by applying DevSecOps to machine learning and making hard easy. My vision is to make analytics data and the resources required for BI & ML easy to access and use with the goal of developing machine learning software that is easy to use and improve productivity.

Mclean, Va, Us

Unleashing the potential of GCP's Vertex AI to provide enterprise level DataOps & MLOp by applying DevSecOps to machine learning and making hard easy. My vision is to make analytics data and the resources required for BI & ML easy to access and use with the goal of developing machine learning software that is easy to use and improve productivity.

Mclean, Va, Us

Mclean, Va, Us

● Analyzed Kaggle data set to predict whether a mobile ad will be clicked or not● Bigger data: Training data set was 6GB (40M rows, 23 categorical columns)● Used Logistic Regression, Decision Tree, Random Forest, & XGBoost● Achieved AUC of 73% with Random Forest and AUC of 70% with Logistic Regression in Python

● Analyzed Kaggle data set composed of transactions made by credit cards in September 2013 by European cardholders which was highly unbalanced, with positive class (frauds) accounting for 0.17% of all transactions● Utilized over- and under-sampling and re-balancing techniques in Python on a Jupyter Notebook● Used machine learning algorithms including Logistic Regression, Random Forest, XGBoost, Support Vector Classifier, and Deep Learning● Achieved F1 Score of 89% with Random Forest and 84% with XGBoost● Achieved Recall of 88% and AUC of 98% with Deep Learning using logit weighting

Philadelphia, Pa, Us

● Re-initiated stalled simulation project to create simulation of Einstein’s Emergency Room to compare existing and proposed new schema● Schema projected to provide the Emergency Room with approximately $3.2 million in additional revenue● Compared new urgent care model to existing model and found underutilization and problem with one of the ESIs, recommending emergency room renovation at a cost of at least approximately $350,000 dollars

Philadelphia, Pa, Us

● Applied PCA and machine learning to produce predictive models with accuracy of 99.8%● Follow-up study on imbalanced data set utilizing over- and under-sampling techniques & Logistic Regression, Random Forest, XGBoost, SVC, & Deep Learning, resulted in F1 Score of 89.4% with target variable .17%● Risk Factors for admission include elderly, female, and obese patients, smokers, and patients on Medicaid● Utilized R (geospatial analytics), SQL (joins), SAS (machine learning), Python (machine learning & NN) & Excel

● Conducted initial screen of all healthcare stocks traded in US using six key financial factors with FactSet● Performed secondary screening based on DuPont analysis using FactSet to find two undervalued stocks● Found two high quality undervalued stocks: TARO & CBPO - TARO (Israel) appreciated 50% within a month of the recommendation - CBPO (China) appreciated 90% within a month of the recommendation

• Analyzed macroeconomic situation to advise portfolio manager on tactical allocations• Created and updated full financial models of companies including scenarios and simulation using Crystal Ball• Used EBITDA Multiple Method & Perpetuity Growth Method to make buy/sell recommendations

Masters, Business Analytics

Finance, Accounting, & Management

Master Of Arts (M.A.), Medical Anthropology (Public Health)

Bachelor Of Science (Bs), Human Biology