Machine Learning Engineer
Current
Argentina
● Managed and analyzed revenue, delivery metrics, and customer behavior for a leading Latam e-commerce site. Built and maintained an ELT data pipeline, processing information from CSV files and APIs. Utilized SQL queries to extract new data, linking tables to analyze relationships like holiday buying patterns. Assessed delivery performance (timeliness and accuracy) and generated data visualizations with Matplotlib and Seaborn to improve business decisions. ● Developed and implemented a credit risk scoring model using machine learning techniques (Decision Tree, XGBoost, LightGBM) to analyze over 350,000 transactions and predict creditworthiness. Preprocessed and visualized data using Python libraries (Scikit-learn, Matplotlib, Seaborn) achieving a training ROC AUC score of 88% and a validation score of 85%. ● Developed and deployed high-performance Machine Learning APIs using Docker. Designed APIs for image classification tasks (e.g., predicting product content, vehicle make/model), achieving accuracy scores exceeding 80%. Utilized data cleaning, visualization, and augmentation techniques to ensure model performance. ● Developed a deep learning model using convolutional neural networks (CNNs) to predict vehicle make and model from unstructured e-commerce images. Achieved over 82% accuracy on a dataset of 196 classes through data cleaning, visualization, augmentation, and fine-grained model training. ● Developed and implemented a sentiment analysis model to categorize movie reviews for a streaming service. Preprocessed and vectorized text data using techniques like Bag-of-Words (BoW), TF-IDF, and word embedding models. Trained and evaluated different machine learning models (Linear Regression, SGD Classifier) to classify positive and negative sentiment. ● Developed and implemented machine learning models (XGBoost and LightGBM) to predict taxi fare and travel time. Developed and implemented high-performance machine learning APIs using Docker Compose.
