Lucas Pereira

I’m a Data Scientist with 5+ years of experience building end-to-end ML systems, from data ingestion and feature engineering to production deployment and monitoring. I specialize in predictive modeling, time series forecasting, NLP, and computer vision, and I bring hands-on MLOps expertise on Azure and Databricks.

At Autoglass, I designed supervised and unsupervised models to support operational and strategic decisions, covering feature engineering, cross-validation, hyperparameter tuning, and error analysis. I built distributed data and ML pipelines with PySpark on Azure Databricks using a medallion lakehouse design, and I implemented MLOps with MLflow (experiment tracking, model registry, and versioning). I also containerized inference with Docker/Kubernetes and automated retraining and deployment through CI/CD (Azure DevOps), including model monitoring and data/concept drift checks.

I’m especially proud of how I move models from notebook to production behind CI/CD pipelines, translating statistical work (EDA, hypothesis testing, inference) into reliable, business-ready outcomes. I’ve delivered forecasting solutions using ARIMA/Prophet and gradient-boosting-based time series models, NLP pipelines for preprocessing and embeddings, and computer vision models with Scikit-learn, TensorFlow, and PyTorch—while also supporting analytics as a Senior Data Analyst by standardizing dashboards and improving query performance.