Data Scientist
Current• Performed Exploratory Data Analysis and developed a machine learning model to predict train faults, alongside building forecasting models using Prophet, resulting in a 25% reduction in damage costs.• Developed an anomaly detection model for airport baggage scanning machines by analyzing sensor and log time series data from 40 units, successfully identifying 30% of failures in the first month of production.• Worked on a Proof of Concept for train fault prediction using high-frequency data, conducting data analysis and building an XGBoost model with a recall of 0.73, presented via a Streamlit application and deployed using Docker and Google Cloud Run.• Implemented a Knowledge Graph for unstructured document data using GPT-3.5 Turbo, improving context-based replies by 30% and enhancing conversational accuracy and user satisfaction metrics.• Addressed production bugs in the model and integrated feedback into the prediction model to improve overall performance.
