At Aliger, I lead R&D projects that combine optical, hyperspectral, and ultrasonic sensors with machine-learning pipelines to detect defects and impurities in industrial streams. I released production-grade APIs and interactive dashboards for real-time quality monitoring.
I also implemented end-to-end model training and inference workflows, containerized with Docker and deployed through GitHub Actions. My work includes five patent applications, two of which have been granted, covering opto-electronic measurement and granular particle inspection.
At EngeData/UFRJ, in collaboration with Petrobras, I developed a web-based platform for sensor reliability that analyzes high-frequency process data and trains anomaly-detection models. I integrated predictive models into Petrobras’ corporate monitoring system to provide real-time confidence indices for inline sensors.
At COPPE/UFRJ, I applied deep learning and ensemble methods to predict lipid nanoparticle production parameters and support inverse-design recommendations. I delivered NanoCore AI, a Python library with stacking ensembles, conformal prediction, and AutoML, accessible through a FastAPI service and React front end.

