I led a project leveraging a machine learning technique called Light Gradient Boosting Machine (LGBM) models to enhance Nike's sales forecasting. The goal was to adapt to Nike's new strategy of focusing on more specific product categories. By training a series of these advanced models, I was able to maintain the accuracy of our sales predictions (measured by MAPE) while significantly increasing the level of detail in the forecasts. This approach not only preserved the quality of our forecasts but also halved the time needed to generate them. The outcome could successfully detail 6-month sales forecasts, in both revenue and units, for all products across 223 US Nike Value stores, utilizing tools such as Snowflake, AWS, S3, SageMaker, and GitHub to streamline the processes.

> Routinely used a variety of machine learning and statistical techniques to reduce fraud, detect anomalies, create custom credit/fraud scores, optimize approval volume, and limit monetary losses> Owned Concora’s credit fraud detection and mitigation strategies with a personal responsibility to reduce monetary losses due to fraud by 5% annually (ID theft and synthetic ID)> Established 2 different vendor partnerships with Concora after investigating investment opportunities with alternative credit data sources, designing implementation, and creating custom score models increasing credit originations by ~4% and ~8% (PySpark, Spark SQL, XGBoost, Matplotlib, MLflow, NumPy)> Managed all credit criteria company-wide. Designed and created the script to process and decision allcredit card applications, ~30K daily (PySpark, Spark SQL)> Set up data stability and performance monitoring dashboards - used to catch and block a major fraudattack that generated a few thousand synthetic ID accounts (Databricks SQL Dashboard, PySpark)> Created a python script to optimize credit card account generation by automatically overlaying models and data attributes, making overlays 5x faster across Risk & Data Science (python, pandas, SQL, scikit-learn, regex, numpy)> Designed and led anomaly detection study for fraud detection using unsupervised learning (PySpark, Spark SQL, NumPy, scikit-learn, scipy, PCA, gaussian mixture, isolation forest, SVM, K-means)> Helped Concora transition from local data servers to Databricks - data verification scripts and dashboards

> Used feature engineering and score overlays to enhance credit scoring model, increasing credit cardoriginations by ~4% (MS SQL, python, decision trees, step-wise regression, scikit-learn, pandas, excel)> Designed and launched A/B/n testing email campaigns to generate accounts and test different productresponse rates on different populations which generated around 10K-20K accounts monthly (MS SQL, excel)> Investigated an outside vendor’s credit dataset for possible implementation. Determined the dataset didnot offer enough additional value on top of existing data for purchase (MS SQL, python, random forest)

> Coach youth soccer players and teams for ages U8-U13> Lead team practices, camps, and technical training sessions

> Lead campus tours - includes anything from intimate one-on-one campus tours to leading groups of 50+ people. > Inform and speak to prospective students about academic departments.> Host live information sessions for prospective families where people can ask questions about the school and I answer them.> Coordinate and schedule visit itineraries for prospective students.> Act as a liaison between students, counselors, and professors.

Summer intern for Carbon Institute in San Diego, CA. Work included critical review and analysis on the emission reports of Indonesia and China, greenhouse gas emission accounting classes, and other small research-based projects.