At DETACOOP, I developed a provision expense forecasting model using Python and machine learning, achieving a MAPE of 2.1% over a 12-month horizon.
I also built automated processes for model training, evaluation, and monitoring, and designed financial data processes that produce analytical datasets and decision-support metrics.
At Constructora Osorio, I designed Python ETL processes to consolidate project information and built Looker Studio dashboards to track operational indicators. I also automated data cleaning, transformation, and validation.
On my projects, I built an Airflow pipeline loading data into BigQuery, developed an energy production forecasting model, and implemented a collaborative-filtering recommendation system using public Amazon datasets. My background in Physical Sciences includes research on complex systems, soliton propagation, and random matrix theory.
