Designed, developed, and configured IaC for 2 ETL pipelines using GCP Airflow, Python and Terraform, automating data ingestion (~100 MB/month) from 3rd-party APIs, removing manual data entry. Delivered 26+ ETL requirements powering LATAM’s Single Source of Truth (SSoT) in BigQuery, integrating 11+ tables and feeding Power BI dashboards for regional leadership. Improved reliability through pipeline monitoring, incident management, root cause analysis, and corrective actions to minimize downtime. Optimized runtime from 25h to 1h using parallelized BigQuery processing and reduced orchestration time from 48 min to 4 min. Led team’s testing strategy and increased Python test coverage sixfold from 2% to 12% using Pytest, improving maintainability and release confidence.

Built ML-powered recommendation engine increasing sales specialist productivity by matching buyer manufacturing needs with relevant supplier products (Gini=0.80). Developed forecasting model for machine production rates (Gini=0.60) using Databricks, integrated with PowerBI dashboards. Trained U.S.-based client on machine learning lifecycle enabling new project launches. Maintained legacy demand forecasting pipeline ensuring production stability and continuous monitoring.

Increased revenue by mitigating credit risk through development of bank’s first ML solutions for credit risk, customer churn, payment collections, and compliance using PySpark, Databricks, AWS. Collaborated in establishing foundational processes and frameworks for data science projects. Improved model training and evaluation by extending in-house ML libraries and leveraging MLflow.