Senior Machine Learning Engineer and Researcher
NAIS S.r.l.Designs production ML and real-time computer vision systems for edge devices, and leads the NAIS Artificial Intelligence team.
About
I work on machine learning systems that have to leave the notebook: cameras, datasets, inference code, edge devices, factory constraints, and the small engineering decisions that decide whether a model is useful in practice.
My main area is visual anomaly detection for industrial inspection. I am interested in methods that are technically sound, but also in the less glamorous parts: memory limits, throughput, calibration, deployment targets, debugging, and long-term maintainability.
Most of my work is in Python and C++, with PyTorch, OpenCV, Halcon, Docker, and embedded or edge-oriented inference stacks. I also keep a few side projects around local agents, Unreal Engine C++, benchmarking, and hardware because they sharpen different parts of the same craft.
Trajectory
Designs production ML and real-time computer vision systems for edge devices, and leads the NAIS Artificial Intelligence team.
Built pharmaceutical visual inspection systems, C++ inference runtimes, and computer vision workflows used under production constraints.
Worked on deep learning and computer vision for out-of-distribution anomaly detection and segmentation in industrial inspection.
Research supervision
Improving PatchCore for Visual Anomaly Detection in Pharmaceutical Product Inspection. Author: Oligert Osmani. Supervisor: Evelina Lamma. Co-supervisors: Niccolò Ferrari and Davide Luisari.
Capabilities