Tiago – Data Science, Machine learning, Python

Creating a document classifier leveraging the power of a pre-trained BERT model fine-tuned on our proprietary collection of documents.

• Crafted the training dataset;
• Developed the training script;
• Conducted in-depth data analysis;
• Trained and evaluated the model;
• Designed and implemented the architecture;
• Established model serving capabilities;
• Constructed Lambdas for backend connectivity.

Collected regulations data about a big industry in the USA and created a question-and-answering system to help recruiters and managers with their questions.

• Conducted comprehensive data collection and analysis;
• Created back-end architecture using Pinecone, AWS lambda, and Langchain;
• Debugged and resolved the application issues;
• Successfully integrated with front-end.

Matchmaking system to offer the best jobs that fit truck drivers signed on our platform. Based on the jobs and driver features, the model could create a list of the jobs that were a better fit for that driver.

• Collected data and analyzed dataset;
• Trained model in AWS Sagemaker environment;
• Deployed model to production;
• Created metrics and dashboards to follow model performance.

Campaign Budget Optimization, or CBO, is an automated ads product that benefits advertisers by distributing the advertising budget for each campaign across the underlying ad groups in an automated manner. The goal of Campaign Budget Optimization is to: Maximize advertiser value, Improve budget utilization, and Simplify the advertiser experience.

• Wrote unit and integration tests
• Refactored the existing back-end
• debugged and resolved the application issues
• Evaluate business metrics after model changes

NLP classification project, using Python and TensorFlow, and MLOps, using TensorFlow Extended (TFX) and TF Serving in a private cloud environment (Docker and Kubernets). This project was designed to process about 30 million requests per month, 24/7, and was integrated with the Brazilian OpenBanking platform, helping clients to organize their expenditures and the company to design better products.

• Collected data, cleaned dataset, and analyzed data;
• Trained model;
• Created API to connect the legacy system and our TF Serving endpoint;
• Deployed models successfully;
• Led a team toward project goals.