Miguel Fernando Uicab Perera

I am a professional with experience in end-to-end product development for Data Science. I have skills in designing tools to predict prices, measure risk, and describe the behavior of phenomena of interest. I am enthusiastic about developing scalable Machine Learning products that help build lasting mutually beneficial relationships.

• Leading participation in the following steps of a ML project lifecycle: Preprocessing, Feature Engineering, Training, Optimization, Deployment and Monitoring.
• Data wrangling. Transform raw data from the data lake into a more usable format for Data Science products.
• Feature Engineering: Construction of new variables based on sources from public institutions or credit information companies.
• Development and deployment of apartment rental price prediction models.
• Development and deployment of risk models to predict late payment or defaulting.
• Use of Time Series and Forecast tools for trend analysis and price updates.
• Use of data mining techniques to extract useful information from events related to the rental market.
• Use of subsampling techniques to deal with unbalanced data.
• Creation of database queries for data analysis and modeling processes.
• Resolution requests from finance, marketing, product and operation areas.
• Support of ad-hoc requirements like small reports, metrics calculation, validation of predictions and statistical analysis.
• Creation of dashboards to summarize and visualize data for decision-making.
• Data storytelling. Creation of presentations to communicate insights and results for the C-level or other teams.
• Constant communication with various internal stakeholders during the development of a Data Science product to successfully meet their specifications.
• Fluid communication with the Technology team to ensure the correct production deployment of Data Science products.

MAIN ACHIEVEMENT: Development of Calculadora Homie Version 5.0, an end-to-end product to predict rental prices in Mexico City. The improvement rate was 25% or more error reduction compared to external competitors, surpassing all of them.

I have two postgraduate degrees related to Statistics and Machine Learning.