Computer vision and machine learning engineer, MSc in Artificial Intelligence. Four years
taking the perception stack for an autonomous robotic car wash from research prototype to a
customer installation running paying traffic: models, robot geometry, edge fleets, and the
tooling that keeps a remote site debuggable.
Technical skills
Machine learning & computer vision
PyTorch, deep learning, image segmentation, 3D reconstruction (RGB-D, NeRF, 3DGS),
photogrammetry, OpenCV, model deployment
Robotics
ROS, path and motion planning, URDF modelling, sensor calibration, embedded C
(FreeRTOS, PULP)
Italian (native), English (full professional proficiency)
Experience
Computer Vision & Machine Learning Engineer
Preen Technologies AG, Zurich· to present
Built and deployed the replacement semantic segmentation pipeline for car-part
identification and dirt estimation, after it outperformed the incumbent model in
production-site validation.
Owned and evolved the RGB-D reconstruction and camera-based vehicle-localisation
stack, validating reconstructed geometry against physical vehicles to millimetre
agreement.
Built and maintained the robot path-generation system across varied vehicle
geometries, resolving production defects and expanding automated regression coverage to
the full production dataset.
Owned a 40+ node Raspberry Pi camera fleet end to end, covering provisioning, AWS
storage and images, and the VPN and remote-access tooling that let the team commission and
debug an international customer site without travelling to it.
Diagnosed cross-domain failures spanning software, networking, electronics and
mechanics, including a fleet-wide camera outage traced to an unexpected fivefold
power-draw anomaly in the enclosure grounding path.
Machine Learning & Embedded Systems Engineer
IDSIA USI-SUPSI, Lugano· to
Developed and deployed CNNs for autonomous nano-drone navigation, estimating the
relative pose and LED state of a peer drone from low-resolution onboard imagery.
Delivered to a severely resource-constrained target, using PyTorch and OpenCV for
training and C on PULP/PMSIS with FreeRTOS on board.
Data Scientist & Software Engineer
Enet Energy SA, Lugano· to
Evaluated a pre-commissioned LSTM-based trading strategy and demonstrated
mathematically that it was not profitable on paper money, preventing live deployment;
ported it from MATLAB to Python and built a real-time MACD system with a custom
grid-search optimiser.
Publications
Ultra-low Power Deep Learning-based Monocular Relative Localization Onboard
Nano-quadrotors. S. Bonato, S. C. Lambertenghi, E. Cereda, A. Giusti, D. Palossi. IEEE ICRA , pp. 3411–3417.
Vision-state Fusion: Improving Deep Neural Networks for Autonomous Robotics.
E. Cereda, S. Bonato, M. Nava, A. Giusti, D. Palossi. Journal of Intelligent & Robotic Systems, 110(58), .
Cyber Security aboard Micro Aerial Vehicles: An OpenTitan-based Visual
Communication Use Case.
M. Ciani, S. Bonato, R. Psiakis, et al. IEEE ISCAS . Also reviewer for IEEE/RSJ IROS .
Education
MSc, Artificial Intelligence
USI Università della Svizzera italiana, Lugano· to