I took over as coordinator at a critical moment: the migration of predictive AI models to a new proprietary planning platform. I led this process with a focus on stability, governance, and alignment between technology and business, without disrupting operations.

Leading a team of ~10 people, I drove the expansion of predictive models to new brands and sales channels, reaching the highest number of models in production in Grupo Boticário’s history. This result was made possible by the team’s technical maturity in MLOps and software engineering, and by a development workflow that allowed us to accelerate deliveries without compromising robustness and scalability.

Beyond technical delivery, I led initiatives to increase model adoption among demand planners, including a forecast reliability indicator and an explainability module that showed the impact of factors such as seasonality, promotions, price elasticity, and cannibalization. I also led a strategic project with an external consultancy, in which our models were validated as a market benchmark.

In managing the team, I prioritized gender diversity (50% women), a collaborative environment, and the genuine development of each person, which was reflected in high engagement scores sustained throughout my time as coordinator, making my management model a benchmark within the department.

As a senior data scientist, my role took on a more strategic and cross-functional character. I actively participated in structuring the corporate data platform on Google Cloud Platform (GCP), contributing to the validation of dozens of tables and translating critical logistics network rules into a democratized and scalable architecture.

I worked closely with the business, applying techniques such as Dynamic Time Warping in an A/B test to evaluate changes in franchisees’ purchasing behavior, an analysis that resulted in a successfully implemented proposal. I was also responsible for rewriting a legacy demand forecasting model in Python with LightGBM, ensuring its integration with the new planning platform being built by the technology team.

It was during this period that I developed the view that data solutions only create value when they are adopted, and I increasingly began to act as a bridge between what technology delivers and what the business actually uses.

I joined Grupo Boticário to develop predictive solutions supporting demand planning, during one of the most challenging periods for any statistical model: the COVID-19 pandemic.

I was responsible for developing a decision tree-based model to predict the probability that a statistical forecast would require manual adjustment by planners. In a scenario of high uncertainty and overloaded teams, the solution became essential for prioritizing which forecasts to review, directly improving the efficiency of the S&OP process.

Beyond modeling, I was responsible for the entire solution lifecycle: deployment, monitoring, and maintenance in production, with governance and traceability. It was my first experience with the full lifecycle of a data product, and it shaped much of who I have become as a professional.

Olist was my definitive entry into data science, and an intense learning ground. I worked on the analytics team focused on the customer relations area, applying predictive models and NLP techniques to real business problems.

I developed the company’s first churn probability model, work that resulted in a scientific publication and showed me, in practice, the full cycle of a data project: from the business question to delivery with measurable impact. I also used natural language processing to analyze customer reviews, generating strategic recommendations that helped improve the company’s reputation on marketplaces such as Mercado Livre and Americanas.

This is where I developed the technical foundation and business vision that I still carry today, and that shaped the professional who joined Grupo Boticário a year later.

At 32, I decided to change careers. I left a solid track record in the Oil & Gas sector to start from scratch in data as an intern, while pursuing a Bachelor’s degree in Mathematics as a second degree. Four months later, I was hired full-time as a Data Analyst.

At Hilab, I took part in implementing Power BI across the company, working from data warehouse modeling on Amazon Redshift and building data ingestion pipelines into Amazon S3 from Azure Cosmos DB, to creating interactive dashboards and reports for different departments. I was also part of a multidisciplinary Growth Hacking committee, focused on gathering and prioritizing hypotheses and designing experiments.

This experience confirmed that the transition was worth every risk. And the systemic view I brought from previous industries has since become one of my main differentiators as a data professional.