Matheus – Python, Big Data, Deep Learning

The cutting-edge forecasting model is designed to predict sales unit forecasts for new products across diverse industries and countries. By integrating state-of-the-art machine learning models and adhering to MLOps best practices, it ensures precision and reliability in predictions.

• Led the technical architecture, development, and implementation of the model pipeline to guarantee its efficiency, reliability, and scalability;
• Directly managed a team member responsible for executing various tasks within the model pipeline, offering guidance, support, and mentorship as necessary;
• Coordinated the integration of our model pipeline with multiple clients, ensuring seamless interaction and alignment with their systems and requirements;
• Committed to achieving low forecasting error rates by consistently refining and optimizing the model pipeline through thorough testing, validation, and performance monitoring.

This project was focused on providing segment pricing for insurance of theft and damage of mobile phones. Based on the user's personal information and history, the pricing in the model is adjusted.

• Spearheaded the model's achievement of a 30% reduction in average premiums while maintaining the same risk level;
• Implemented the pricing model using Tweedie regression;
• Utilized Docker and API integration for model deployment;
• Oversaw model performance monitoring;
• Orchestrated project development;
• Ensured alignment with business stakeholders.

Machine learning model tailored to assess insurance claims for automatic approval based on historical data. This innovative model drastically reduced claim payment processing time from a potential maximum of 14 days to a mere 2 hours. Also, it successfully automated the payment of 30% of claims, resulting in significant savings in human analysis time. In essence, the model revolutionized the efficiency and speed of claim processing while enhancing cost-effectiveness through automation.

• Developed the model for claims evaluation, ensuring accuracy and efficiency in decision-making processes;
• Created comprehensive unit tests to validate the functionality and reliability of the model;
• Deployed the model for claims evaluation via API integration and Docker containerization, ensuring seamless integration;
• Implemented monitoring mechanisms to assess the model's performance in evaluating claims;
• Managed the project lifecycle, including planning, execution, and monitoring of tasks and milestones;
• Collaborated closely with business stakeholders to understand requirements and ensured alignment between model development efforts and business objectives.

This is an intent recognition model based on WhatsApp chat for detecting customer friction.

• Developed and deployed machine learning models;
• Monitored model performance;
• Oversaw model project management.