Helped lenders in Africa, Asia and the Americas make better decisions to serve people with no credit history, mainly micro-entrepreneurs. To do this, I built customer-specific and regional credit risk models with various alternative data sources, including digital psychometric questionnaires, metadata, geolocation (GIS) and digital presence.

• Feature engineering.
• Development, training and evaluation of predictive models.
• Implementation of statistical and machine learning models.
• Model reporting and model performance evaluation.
• R & D tasks: methods for unbalanced data problems, model interpretability, and inclusion of new data sources.

Software: R, Python, SQL, PostgreSQL, MongoDB, AWS, IBM Watson

Helped Peruvian retail and financial companies understand their customers and find better opportunities by exploiting big data and especially geolocation information (GIS).

• Participation in the definition of the problem and exploration of the client company’s needs.
• Technical support in the formulation of proposals.
• GIS data analysis. HeatMaps, and Trade Area.
• Site Selection based on spatial data analysis to open new Banking Agencies and Retail Stores.
• Resource Allocation improvements with Geospatial Analysis.
• Supervised and Unsupervised models for Client Segmentation, Sales Potential, Income Prediction, and Customer Churn.

Software: R, Azure ML, PostgreSQL, SQL, ArcGIS, QGIS

Contributed to the analysis, evaluation, and monitoring of psychometric credit risk models used in African banks serving micro-entrepreneurs in Kenya, Nigeria, South Africa, Tanzania, and Botswana.

• Various descriptive data analysis, and reporting tasks,
• Model reporting.
• Support in data preparation: automation of data extraction pipeline, cleaning, and feature engineering.

Software: Stata, R, VBA Excel