Bruno Silva Sette

Profile

I'm a developer with 13 years of experience, a background in Data Science, and a Master's degree in Machine Learning. Highly motivated, constantly improving my skills, and pursuing professional growth.

Work History

Data Scientist Intern | AMERICANAS S.A

2019 - 2021

• Developed AutoML solutions using the Apache Marvin AI framework, streamlining the automated search for optimal machine learning models.
• Conducted in-depth research on Automated Machine Learning (AutoML) and Neural Architecture Search (NAS), focusing on improving model efficiency and accuracy.

Data Scientist | AMERICANAS S.A

2021 - 2023

• Trained deep learning models using MLOps tools such as Docker, Kubernetes, and CI/CD on large-scale datasets (over 10 million product titles and images), orchestrated with Airflow and Vertex AI Jobs, and processed using Dataflow jobs.
• Demonstrated experience working in multi-cloud environments, leveraging both Google Cloud and AWS.
• Applied bottom-up approaches (H-DBSCAN) for hierarchical product classification, identifying and removing outliers from the training set of hierarchical deep learning models.
• Researched hierarchical multi-modal classification techniques (combining text and images) for e-commerce product categorization.

Data Scientist | SOFTPLAN UNJ

2023 - 2025

• Deployed NLP machine learning models to production using MLOps tools such as Docker, Kubernetes, and CI/CD, ensuring scalable, reliable, and automated model serving pipelines on AWS ECS and Kubernetes clusters.
• Implemented ETL processes for SQL Server, PostgreSQL, and Oracle SQL databases from the Brazilian judicial system, handling servers with different versions and distributions located across the country.
• Created and deployed an OCR PDF API using AWS Step Functions and AWS Lambda with AWS Textract and a version with Redis, RabbitMQ and Celery for clients environments.
• Applied hierarchical clustering techniques for semi-supervised learning to build a taxonomy of legal processes for state Prosecutor Offices.
• Led and mentored a team of three junior data scientists, sharing both academic knowledge of machine learning and practical expertise in NLP, OCR, and ETL projects involving judicial data from the Brazilian judicial system.
• Developed and maintained a fork of the open-source Open WebUI project, named Open WebUI Iara, for the Brazilian Public Prosecutor’s Office (Ministério Público – MP). This version features asynchronous document processing with Celery, RabbitMQ, and Redis as the result backend, significantly enhancing scalability and efficiency when managing large volumes of data. Integration with Weaviate was implemented via GraphQL on FastAPI routes, ensuring robust and scalable semantic data retrieval.

Data Scientist | Solides

2026 - Present

• Developed a customer churn prediction agent for Solides by building automated training and inference pipelines on Vertex AI, standardizing the machine learning lifecycle from model development to production scoring.
• Engineered data pipelines to publish model outputs, customer features, and health indicators into BigQuery, enabling reliable monitoring and downstream analytics for retention initiatives.
• Built dashboards to monitor customer health across the company, giving business teams visibility into risk signals, engagement metrics, and retention opportunities.
• Developed a next-best-offer pipeline based on association rule statistics, generating product offer recommendations that the commercial team can consume directly through its CRM.
• Developed an AI Agent using the Google Gen AI SDK and Model Context Protocol (MCP) to interface with BigQuery, enabling autonomous consumption and reasoning over customer Health Score metrics to drive proactive retention insights.
• Led AI maturity initiatives by organizing the company-wide Gemini Enterprise trial, governing license allocation, connector setup, usage monitoring, and support for teams adopting AI across the organization.

Education

Computer Network Technician | Serviço Nacional de Aprendizagem Industrial

2012 - 2014

• Recognized as the top student among all technical courses in the graduating class of the second semester of 2014.

Bachelor of Computer Science | Universidade Federal de Mato Grosso

2015 - 2019

• Active participant in the CNPQ Scientific Initiation Program, focusing on NLP and hierarchical text classification for industrial patents from the World Intellectual Property Organization (WIPO).

Master of Computer Science (M.Sc.) in Machine Learning | Universidade Federal de São Carlos

2019 - 2021

• Conducted a literature review and published advancements in Neural Architecture Search (NAS) through an ensemble approach to search spaces and model combination, demonstrating that dividing the search space into different network configurations can improve final classification metrics.

PhD Candidate in Computer Science | Universidade Federal de São Carlos

2021 - 2027 (Ongoing)

• Researching hierarchical multi-label classification techniques and developing a framework to advance hierarchical multi-label categorization metrics using deep learning with PyTorch.
• Current work focuses on metric learning to improve hierarchical multi-label metrics based on the research hypothesis.

Contact

• Email: brunosilvasette@gmail.com
• Phone: (11) 93536-4670