Independent validation of Zetyra statistical calculators — 629 tests across GSD, CUPED, Bayesian, SSR, RAR, and Master Protocol (basket, umbrella, platform) calculators. Includes I-SPY 2, STAMPEDE, and REMAP-CAP trial replications.

Machine learning pipeline for predicting used car prices using Kaggle’s Playground S4E9 dataset. Includes data cleaning, EDA, feature engineering, statistical analysis, and advanced regression models (LightGBM, XGBoost, CatBoost). Modular notebooks, visuals, and documentation included.

Teste A/B com Python para validação estatística de um sistema de recomendação em e-commerce.

Secure Neural Network Model Watermarking

Calculates the potential maximum daily loss for an investment portfolio using three different financial risk methods (Value-at-Risk). The system checks these methods are accurate using statistical backtesting, helping set safe risk limits and meet financial rules.

End-to-End Python framework implementing bias-adjusted LLM agents for human-like decision-making in economic games (Kitadai et al., 2025). Features persona-conditioned agent populations using Econographics data, multi-provider API integration, and Wasserstein distance validation against empirical benchmarks.

Research-grade implementation of Bidirectional ALT for shortest path problems — achieving up to 8× speedups on structured graphs with full statistical validation.