Tania España

I'm a Data Scientist at Walmart Central America, where I develop machine learning models that feed directly into operational decisions across the region. My largest build there is a demand forecasting system combining LightGBM models with M/M/1/K queuing theory, predicting hourly checkout demand, service speed and required staffing up to 180 days ahead for every market in Central America, with automated recommendations that keep utilization within target. I work across the full modeling cycle — extracting and structuring data in SQL and BigQuery, exploratory analysis, feature engineering and preprocessing, development in Python with scikit-learn, hyperparameter tuning and validation — and I defined the data quality and governance standards the analytics function runs on. My wider portfolio covers supervised and unsupervised learning across several problem types: churn prediction on strongly imbalanced data, corrected with over- and undersampling and tuned to prioritize recall of the minority class; customer retention modeling with Random Forest, deployed alongside a decision dashboard; hourly demand forecasting with an optimized neural network; behavioral classification for plan recommendation; investment selection under uncertainty using regression with bootstrapped risk analysis; NLP review classification with BERT and a full tokenization, lemmatization and TF-IDF pipeline; and computer vision with CNN and ResNet50 transfer learning for age estimation. I came to data science from Industrial Engineering and several years in IT project delivery, data warehouse implementation and automation — a background that taught me how models actually get adopted rather than shelved. I'm most interested in roles where models are deployed into real decisions rather than delivered as reports.