Patrick Alves

I’m a Machine Learning Engineer with 5+ years of experience building and deploying machine learning solutions for real-world applications. I led end-to-end ML pipelines in microservices architectures, reducing processing time from days to hours while optimizing cloud costs.

At Margia, I built and deployed demand forecasting and pricing models to improve pricing decisions and inventory planning. I created scalable data pipelines for large-volume ingestion and standardization, and I managed experiments and model versions to improve organization, traceability, and development speed.

I also develop AI-powered tools that analyze data and suggest actions, making teams faster and more informed. I apply model explainability techniques to make results more transparent, increasing trust and adoption among stakeholders, and I automate CI/CD with GitHub Actions to improve release reliability.

Previously as a Machine Learning Researcher (H.IAAC), I engineered pipelines for large-scale sensor datasets and built time series features using statistical and frequency-domain methods (mean, variance, correlation, FFT). I trained supervised and unsupervised models for human activity recognition (including CNNs), used t-SNE and UMAP for exploratory pattern discovery, and applied SHAP and LIME for Explainable AI. I’ve also published scientific work in Nature Scientific Data and presented at BRACIS.