Jorge – Python, SQL, PySpark

A consultancy firm that works with data from the gambling industry.

• Created a system that allowed operators to communicate with players likely to develop problematic behavior
• Increased the lifetime value of players with a recommendation system
• Developed a centralized knowledge platform
• Deployed a real-time pipeline for in-game model intervention
• Clustered models to detect players that were likely to develop an addiction
• Developed the productions pipeline to serve as a powerBI dashboard where the results of the previous models alongside other KPIs were displayed
• Did the development and deployment of a live churn model to extend time on session of a player by predicting the exit spin ahead of time.
• Did code reviews.
• Did the code refactoring of junior Data Scientists.
• Built recommendations systems for sports betting

Jorge was part of a consultancy that was doing a project for Vodafone, and the main goal of this project was to migrate all the machine models that were deployed in local servers to the cloud.

• Did the deployment of Machine learning models on the cloud.
• Did the development of unit tests.
• Tested and calibrated machine learning models.
• Did code reviewing.

Jorge was in charge of testing and developing a new machine learning platform to be used by data scientists. In order to test this platform he developed two models:

• A graph-based approach to cross-device identification of users.
• A graph approach to model multi-channel attribution channels on marketing.

• Developed AWS machine learning platform built on microservices
• Created graph database for cross-device identification
• Developed a machine learning environment where data scientists could easily deploy machine learning models
• Developed a testing environment for machine learning
• Developed a cross-device model to be deployed and to test the platform
• Participated in dev ops development of a machine learning platform
• Developed a production pipeline for a team of Data Scientists to be able to deploy, maintain and improve machine learning models

A multi-channel attribution model to optimize company's marketing budget and also a model to segment their audience.

• Optimized a budget allocation per channel to improve conversions and clicks
• Used Markov chains approach to model the importance of a channel to conversion
• Optimized algorithm to allocate budget based on the information on the Markov model.
• Used the Bayesian survival model to cluster users according to their business value.
• Deployed machine learning models on the cloud with an API service.

A media agency where most of its produced content is published on social media platforms. Jorge was hired to build a daily data gathering system of that content from different social media platforms.

• Created a complete system data that grabbed data from social media platforms.
• Developed the NLP sentiment analysis models to help the marketing team make decisions.
• Built Google app engine application using Python that would capture data using different API connections and stores it on a MySQL database
• Did training set construction tool built-in Django
• Set natural language processing model for sentiment analysis
• Did D3.js dashboards for PPC performances.

The platform to help developers on improving games under development via users feedback

• Built an automated system to classify text on a gamer's feedback platform
• Used C# Natural language processing max entropy approach to approve or reject user’s feedback regarding the content
• Supported vector machine for sentiment analysis.
• Maped and reduced jobs for ETL on the Azure platform to get Microsoft Paint telemetry data
• Built the new version of Microsoft Paint by analyzing telemetry data