Lead Data Scientist2016 - PRESENTCrunchFlow (via Toptal)
Technologies: Python, Azure, Machine Learning
- Created human resource analytics models.
- Forecasted employee churn.
- Forecasting candidates' KPIs using machine learning.
- Optimized resource allocation.
- Created a different kind of API to allow usage of machine learning modules.
Lead Data Scientist2015 - PRESENTEasy Solutions, Inc.
Technologies: Python, R, Sklearn, Big Data, NoSQL
- Managed the data science team.
- Developed machine learning models for information security.
Professor of the Master in Analytics2016 - 2016Universidad de los Andes
Technologies: Machine Learning, Statistics, Data Science, Natural Language Processing
- Oversaw courses in natural language processing, big data, and machine learning.
PhD Researcher2012 - 2015University of Luxembourg
Technologies: Python, R, Sklearn, Spark, SQL
- Developed example-dependent cost-sensitive classification techniques.
- Created a machine learning technique tailor-made for credit card fraud detection.
- Applied cost-sensitive predictive modeling to a variety of real-world applications such as credit card fraud detection, credit scoring, churn modeling, and direct marketing.
Fraud Data Scientist2012 - 2015SIX Financial Services
Technologies: Python, Sklearn, R, SQL, Oracle
- Developed intelligent reporting to support the card management team.
- Implemented advanced cost-sensitive classification credit card fraud detection models.
Data Scientist2010 - 2012Scotia Bank/Colpatria Bank
Technologies: SAS, R, VBA, MATLAB, PHP, SQL
- Implemented genetic algorithm and particle swarm optimization models in SAS for selecting the best architecture of a multi-layer perceptron neural network, and for selecting the variables that maximize the KS statistic in a logistic regression model.
- Created different cluster analyses for the risk and marketing areas, for clients segmentation and model segmentation, among others.
Statistical Models Analyst2008 - 2010GE Money/Colpatria Bank
Technologies: SAS, SPSS, R, SQL, VBA, PHP, MATLAB
- Developed acquisition and behavior scorecards for calculating clients' probability of default, using logistic regression, CHAID decision trees for variables binning, binary genetic algorithm optimization for variable selection, and multi-layer perceptron neural networks.
- Created a constraint optimization algorithm for assigning collection treatments to bank clients, using the probability of a client of falling in next bucket as an input, the expected response per client per treatment, total balance, and treatments costs.
Six Sigma Intern2006 - 2008The Dow Chemical Company
Technologies: Oracle, VBA, SAS
- Developed reports for the commercial and marketing areas.
- Created GARCH and ARIMAX models for forecasting raw materials prices.
- Responsible for a Six Sigma project for time cycle reduction on international orders. The result was the building of a new warehouse on a Colombian free trade zone.
- Developed several marketing research projects for plastics, construction, and chemical departments.