João Victor Castro de Oliveira

• Led a project to structure data for campaigns, levers, and retention vouchers.
• Designed analyses to optimize voucher pricing, maximizing retention and EBITDA impact.
• Improved and maintained a churn prediction model using XGBoost.
• Built dashboards and reports in Tableau to provide visibility into initiatives and support decision-making.
• Supported the architecture of an internal chatbot using LLMs to answer FAQs and enable data-informed decisions.

• Structured the CRM data architecture by unifying multiple sources into the Data Lake.
• Managed data ingestion through PostgreSQL, SFTP, Salesforce, Databricks, and others.
• Developed ingestion pipelines using PySpark and orchestrated workflows with Airflow, following Delta Lake architecture.
• Led the entire data product cycle: ideation, MVP, and handoff.
• Proposed technical solutions that reduced costs and improved efficiency.
• A dedicated data team was formed after the MVP to support the new structure.

• Performed segmentation and campaign analysis in CRM.
• Versioned segmentation notebooks using GitHub and CI/CD.
• Automated ETL workflows with Airflow and integrated with Salesforce Marketing Cloud.
• Defined personalized customer journeys and KPIs.

• Founded and led the data area.
• Built models using neural networks, KNN, SVM, QDA, KMeans.
• Created dashboards with Amazon QuickSight and customer consumption profiles for B2B and retail.
• Developed a full NLP pipeline on AWS for sentiment extraction from messages.
• Achieved a 25% increase in lead conversion through propensity models.
• Delivered technical presentations to clients focused on business value.

• Classified throat diseases using neural networks, achieving 98% accuracy.
• Extracted signal features using Poincaré section analysis.
• Tutored STEM subjects (Analytical Geometry, Linear Algebra, Calculus).