Led the development of a Python-based NLP query and prediction application using HuggingFace’s BERT models in Streamlit to analyze satellite testing anomaly reports, which significantly improved similar report retrievability and root cause prediction accuracy.Assumed data science leadership of a supply chain cost and lead time prediction application using categorical embeddings, clustering, and XGBoost. Collaborated closely with the client on UI design and led user acceptance testing (UAT) to ensure smooth system adoption and user satisfaction.Automated the data and modeling workflows for an autoencoder-based normal behavior modeling (NBM) application using industrial equipment time series sensor data. Optimized the process from data ingestion to model training to visualization of results, predicting maintenance needs and preventing equipment breakdowns through proactive anomaly detection.

Researched the application of machine learning to the Sofia SARS Antigen Fluorescent Immunoassay qualitative (i.e., positive/negative results) test images to extract quantitative (numerical) antigen concentration predictions.Designed and built machine learning convolutional neural networks using Python, Tensorflow, and Keras, achieving a 60% reduction in coefficient of variation (prediction error) compared to a four-parameter logistic analytic method.

Automated database creation and queries using Python, SQL, and MS Access to generate complex combinations of mechanical, thermal, aerodynamic, and vibrational loads for application to finite element models of aircraft structures.Performed structural analysis for metallic and composite parts on commercial aircraft nacelles and authored and reviewed analysis reports for FAA Certification on Airbus A320neo, A220, Embraer E190/195-E2, and E175-E2.Guided and mentored remote stress team including detailed formal review of technical reports.