Open to opportunities

Gabriel Aguiar

@gabrielaguiar

Message

Senior Data Scientist turning machine learning into production decisions for regulated, high-stakes business outcomes.

Brazil

Message

What I'm looking for

I want to build and deploy end-to-end ML systems and decision pipelines with strong MLOps ownership, stakeholder-ready explainability, and measurable business impact—especially in data-rich, regulated environments.

I’m a Senior Data Scientist with 3+ years building machine learning systems that turn data into actionable intelligence for high-stakes, regulated environments. I focus on translating complex predictive models into scalable analytical solutions that stakeholders can trust and act on.

At Banco do Brasil, I led a fund cash forecasting and allocation initiative using PySpark, DB2/Hive SQL, and time-series ensemble modeling (XGBoost, LightGBM, Prophet). The production decision pipeline supports an estimated ~140M BRL projected annual yield opportunity, and it’s deployed and monitored at scale.

I also applied SHAP-based explainability to communicate model behavior to product owners and business stakeholders, and I built decision logic that balances operational penalties. In parallel, I developed and deployed a rule-based decision pipeline on facial biometric data with ~8M BRL annual impact, from design through production monitoring.

I standardize delivery with Kedro-based ML pipelines (testing, versioning, reproducible patterns) and I’ve strengthened adoption across the Analytics Intelligence unit. Beyond production work, I teach MLOps and applied ML—delivering Kubeflow and AWS SageMaker training, mentoring 500+ students—and I’ve built analytics products like Power BI retention dashboards that increased retention by ~20%.

Experience

Work history, roles, and key accomplishments

Current

Data Scientist

Current

Banco Do Brasil

Jul 2025 - Present (11 months)

Led a fund cash forecasting and allocation project using PySpark and time-series/ML models, targeting ~140M BRL projected annual yield improvement and delivering a production decision pipeline with ongoing monitoring. Built stakeholder-ready SHAP explainability and deployed a rule-based facial biometric decision pipeline with ~8M BRL estimated annual impact.

Pyspark Python XGBoost LightGBM Prophet SHAP Explainability Kedro Time Series Models

Current

Data Scientist Instructor

Current

TripleTen Brasil

Apr 2025 - Present (1 year 2 months)

Designed and delivered 10+ applied ML projects covering time series, classification, and optimization scenarios. Mentored 500+ students through complete analytical cycles using Python and SQL, with an emphasis on analytical reasoning and clear communication for non-technical audiences.

Machine Learning Python SQL Time Series Modeling Classification Optimization

MLOps Instructor

Refatorando

Feb 2026 - Mar 2026 (1 month)

Delivered a hands-on MLOps course covering Kubeflow and AWS SageMaker, focusing on model deployment and pipeline orchestration for scalable ML workflows. Produced practical end-to-end implementation materials based on real-world ML system needs.

Kubeflow AWS Sagemaker MLOps Technical Instruction

Data Scientist

BITKA Analytics

Aug 2023 - Jul 2025 (1 year 11 months)

Refactored a notebook into a parameterized Kedro MLOps pipeline, reducing runtime by 90% (~20 min to ~2 min) and improving testability and reproducibility. Integrated three predictive models with nonlinear optimization (~40 operational constraints) to generate actionable setpoints and shipped a Streamlit app for scenario validation by operations teams.

Kedro Python Pyspark Streamlit predictive modeling MLOps Testing & Versioning

Data Analyst

Injet Care

Apr 2023 - Mar 2025 (1 year 11 months)

Increased customer retention by ~20% by building Power BI dashboards and performing customer segmentation to support targeted retention campaigns. Led customer grouping and procedure recommendation analyses to support early-stage churn modeling and retention targeting workflows.

Power BI Customer Segmentation Retention Analytics Churn Modeling Data Visualization Procedure Recommendation