At Monks, I built Gemini-powered comment-tagging tools that turn tens of thousands of unstructured comments into labeled data, saving about 38 hours of manual work per 10,000 comments.
I also built and calibrated Bayesian models with uncertainty quantification on multivariate time-series data, using posterior estimates to support resource-allocation optimization. I collaborated on migrating statistical engines and charts from Streamlit to Flask.
For my MSc thesis at Universidad Austral, I built a Python pipeline to cluster F1 drivers by driving style using FastF1 telemetry. I engineered braking, apex, and exit features, then applied K-Means++ and PCA and built a Streamlit dashboard to explore driver lap profiles.
My other work includes time-series forecasting and scenario simulation at Market Fusion Analytics, plus a LightGBM bank-churn model developed for MSc coursework. I have an actuarial and statistics background and am pursuing an MSc in Data Science.
