CV

Robot Learning Researcher

Araya Inc.

• Designed and executed a real-robot evaluation of six Vision-Language-Action (VLA) models, collecting tens of hours of teleoperation data and training and deploying every variant, resulting in ∼5,000 rollouts.
• Open sourcing all the policy rollouts across six models and spanning more than 20 robot hours. Top contributor at oopsie-data.com.
• Developed statistical analyses of task performance and trajectory quality beyond standard success-rate metrics.
• Built imitation-learning pipelines for robot manipulation, fine-tuning and deploying image- and point-cloud-based policies on physical robots across multiple projects.
• Open-sourced a VLA batched inference and kv-cache implementation that improved throughput by up to 40× for the DSRL algorithm.
• Implemented operational-space and Control Barrier Function-based controllers and teleoperation interfaces, enabling safe data collection and repeatable deployment of learned policies on physical robots.
• Developed custom MuJoCo simulation environments and robot-control interfaces used throughout a user study with 20+ participants.

Visiting Research Student

Tohoku University

• Developed compliant control methods for robotic assembly tasks and a Model Predictive Control scheme for collision avoidance.
• Built a dual-arm solar-panel assembly pipeline as an MSc thesis project, leading to a peer-reviewed publication at SII 2025; project developed in ROS.

MSc

Sapienza University of Rome

Artificial Intelligence and Robotics

• Thesis: Autonomous Assembly of Solar Panels by Dual Robot Arms in the Remote Space.
• Supervisors: Prof. A. De Luca and Prof. K. Yoshida.

BSc

University of Campania Luigi Vanvitelli

Electronic and Computer Engineering

• Thesis: Object Detection with Neural Networks for Robotics Applications: Training Methods Compared.
• Supervisor: Prof. C. Natale.