At TripleTen, I designed end-to-end analytical workflows across 15 projects, using Python to explore data and shape recommendations for small and medium-sized businesses. I also developed classification models for customer churn, reaching 86.7% AUC-ROC.
For a simulated $100M USD investment project, I used bootstrapping and cross-validation to reduce the estimated probability of capital loss to 0.50%. My projects also include a Beta Bank churn model that achieved a 64.41% F1 score and identified customer age and number of products as key variables.
My background in geology and mining engineering gives me a quantitative approach to evaluating projects under uncertainty. Before moving into data science, I operated a family food business and analyzed demand patterns to reduce operating losses by 20%.

