Bank-style Credit Risk Scorecard using Logistic Regression, IFRS-9 Expected Credit Loss, and an Interactive Streamlit Risk Dashboard for loan default prediction.

Projeto da API do primeiro semestre de 2026

A Python library providing elegant, type-safe primitives for building the kinds credit models that typically force teams to reach for Excel.

A dirty-work toolbox for data analyses about fixed income securities, developed by only me (not the institute) as an intern data analyst.

Production-ready FastAPI service that converts Credit Reports (PDF format) into structured JSON data using CreditGraph AI patterns with automatic PII scrubbing for data privacy.

Coding assignments of the "Machine Learning in Finance & Insurance" course at ETH Zürich (Fall 2024).

The Credit Product Recommendation Engine

Predict financial risk using behavioral and demographic data from the 2021 FinAccess Household Survey (KNBS). Built with Streamlit and XGBoost.

End-to-end AI Fraud Detection & Transaction Monitoring project using SQL, Python, ML models, SHAP explainability, and FastAPI integration.

Simulação de concessão de crédito em uma instituição financeira. Analisa variáveis como renda, idade e histórico de inadimplência para entender padrões de aprovação e reprovação, gerando insights estratégicos para decisões baseadas em dados.

A predictive credit scoring system using alternative behavioral and demographic data from the 2021 FinAccess Survey to assess household loan default risk in Kenya.

Plim AI First

Ecosistema analítico de riesgo de crédito de punta a punta. Integra metodologías actuariales y estándares de Basilea III para transformar datos crudos en métricas de solvencia financiera y gestión de capital.

Data preparation, predictive modeling and classification, conclusions and recommendations. Preparation and modeling preformed in Python. Work in progress.

Predicting credit card default using machine learning (Logistic Regression, Random Forest, XGBoost) on the UCI Taiwan Credit Card dataset. Covers EDA, feature engineering, class imbalance handling, and model evaluation with AUC-ROC and SHAP explainability.

Deep learning-based credit risk prediction system using neural networks to assess loan default probability. Implements multiple ML/DL models with feature engineering and class imbalance handling. Built with PyTorch for accurate credit scoring and risk assessment.

DEBT DESTROYER - Avalanche Method Tracker

This project analyzes credit card customer & transaction data to uncover key business insights.

AI-Powered Alternative Credit Scoring for 2+ Billion Unbanked People Worldwide

A simple, interpretable credit approval model built on HMDA data using decision trees