International ice dancer representing Brazil. Co-Founder & CTO building AI automation for enterprise. Data science student at LSE. Operating at the intersection of elite athletics, machine learning, and venture-building. The kind of person who qualifies for ISU Championships in the morning and deploys production AI systems in the afternoon.
Competing internationally while building a company from nothing requires the same skill: relentless resourcefulness in the face of zero infrastructure.
Coursework spanning machine learning, financial statistics, Bayesian inference, and time series modeling. Balancing a rigorous academic program with international athletic competition and running a revenue-generating startup.
Structured comparison of linear, regularised, and tree-based models (Logistic, Ridge, LASSO, Decision Tree, Random Forest) for estimating 12-month recession probability from macroeconomic indicators. Designed a business-cycle cross-validation framework to prevent look-ahead bias and evaluated models on AUC, Brier score, and log-loss under class imbalance.
Investigating how much predictive power bookmaker-implied probabilities add when appended to feature-based ML models predicting Premier League match outcomes. Built baseline models using performance features (xG, shots, possession) over 10 seasons of data, then quantified the marginal information gain from betting market odds using log-loss and Brier score differentials with temporal train-validation-test splitting.