> ABOUT.TXT
I'm a data scientist at Stone, one of Brazil's largest payments companies, where I build credit risk, behavior scoring, and propensity models that help decide how we extend credit to merchants. My work runs the full lifecycle — feature engineering, master-table construction, model training and evaluation, and production monitoring with retraining — mostly in Databricks with PySpark and SQL. Beyond credit modeling, my background spans the wider ML landscape. I've worked across scikit-learn, TensorFlow, and Keras, with hands-on experience in computer vision using OpenCV, and I'm currently deepening that foundation through an MSc in Computer Science focused on Artificial Intelligence. Outside of work I'm a persistent, curious learner: I read widely, study languages (fluent in English and German alongside Portuguese), and shoot analog photography.
> EXPERIENCE_LOG
- Build and productionize credit-scoring models, from feature engineering and master-table construction through training, evaluation (AUC, KS, PSI), and deployment in Databricks with PySpark and SQL.
- Monitor production credit models for drift and stability, retraining as performance degrades.
- Develop propensity and conversion models combining internal and external credit data to support credit-offer decisions.
- Optimize large-scale Spark/Databricks pipelines for performance and reliability.
- Partner cross-functionally with credit, policy, MLOps, and data-engineering stakeholders to scope and ship models.
- Built statistical and machine-learning models for credit risk and provisioning.
- Served as technical reference for compliance with CMN Resolution 4.966/21 (IFRS 9), directly shaping regulatory compliance and model performance.
- Led data science teams across machine-learning projects in the healthcare, energy, and steel sectors.
- Delivered computer-vision classification tools, time-series predictive models, and large-scale data processing and visualization solutions for clients.
Developed machine-learning solutions for client companies using Python, OpenCV, PySpark, scikit-learn, and SQL.
> EDUCATION.DAT
> RESEARCH_LOG
Master's research focused on artificial intelligence. Project on using GANs for pseudo-colorization of medical images to analyze cancer progression.
Built a tumor-detection algorithm in Python using random forests on the lab's cancer-biopsy database, reaching 98% accuracy on healthy-vs-tumor classification and 70% across six tumor types.
Designed and ran an experiment on how students' perception of their academic success relative to peers affected their grades over a semester, analyzing results in R.
Study group on evolutionary game theory, statistics, and time-series analysis with R.