Project Portfolio

Commodity Forecasting Project

Multinational Commodities

Objective

Time series forecasting for commodities.

Achievements

Reduction in expenses and increase in profits with up to 3-month price forecasting for various commodities used in the base product manufacturing process.

Tools

Azure Databricks Tensorflow Keras Scikit-Learn SQL

AI Model Governance Project

Multinational Food

Objective

Create and manage pipelines for Machine Learning model workflows in production.

Achievements

Reduction in productization time through automation and reduction in operational costs through workflow standardization.

Tools

Azure Databricks MLFlow MLOps CI/CD Python

Inventory Optimization Project

Multinational Food

Objective

Create models to forecast sales of products nearing their expiration date.

Achievements

Reduction of product waste by indicating expiration and stock renewal for retailers, generating savings of thousands of reais.

Tools

Azure Databricks MLOps Tensorflow Keras Scikit-Learn

Retail Recommendation Project

Multinational Beverages

Objective

Recommend products based on customer history and company launches.

Achievements

Reduction in customer purchase time with automatic item suggestions, transforming the salesperson into a strategic consultant and increasing visit efficiency.

Tools

Azure Databricks Tensorflow Keras SQL

Legal Classification Project

Startup Judiciary

Objective

Create a model to classify legal documents.

Achievements

Reduction in time and savings in processing costs with automated analysis.

Tools

LLM Ollama MLFlow MLOps Prefect AWS S3 Docker

Logistics Route Optimization Project

National Logistics

Objective

Perform customer clustering to optimize delivery based on route and merchandise type.

Achievements

Cost reduction for the company by grouping customers with similar loads on optimized routes, using the ideal vehicle for each delivery.

Tools

QlikSense Scikit-Learn ETL SQL Python

Banking Fraud Detection Project

National Financial

Objective

Create a model to prevent fraud carried out through photos in banking applications.

Achievements

Reduction in operational costs, greater customer acceptance and reduction in fines/lawsuits for fraud.

Tools

Pytorch Tensorflow OpenCV SQL

Real-Time Object Detection Project

National Engineering

Objective

Create a high-performance model to detect custom objects through cameras in real time.

Achievements

Reduction in alert time, enabling 24/7 operations and optimization of the operators team.

Tools

Pytorch OpenCV YOLO Docker Computer Vision