Hi, I'm Alessandro De Vidi 🚀

Mobile app for digital asset management and inheritance planning. Built with Flutter and Firebase, featuring secure data management, post-mortem access protocols, and comprehensive succession planning tools. Simplifying estate management for the digital age.

Zero-shot approach to improve Vision-Language Model performance on Oxford Flowers102 dataset using LLM-generated prompts crafted from runtime knowledge base. Leveraged web search for visually rich textual information and image augmentations via Zero-1-to-3 diffusion model for enhanced similarity estimation.

Distributed multi-agent AI system revolutionizing home repair assistance through intelligent problem diagnosis (87.5% accuracy), professional matching with Trust-Based algorithm, and automated service coordination. Built on microservices architecture with Agent-to-Agent communication protocols.

Complete CV pipeline for 3D human pose estimation using multiview cameras. Triangulated 2D keypoints to recover basketball poses in 3D with 18mm MPJPE accuracy against motion capture ground truth. Calibrated 8-camera rig with automated lens-distortion correction.

Co-developed the vision system of a robotic beer dispensing system for RICAP @ CIIRC Prague / Expo 2025. Integrated KUKA arms, neural networks, and real-time sensor feedback with focus on optimization and precision.

Finalist (3rd place) in "100 Software Projects for the City" challenge. Developed a full-stack web service to improve public utility access and citizen interaction with Municipality of Trento.

Bachelor's thesis: developed new strategies with ITI-GEN and Large Language Models to mitigate bias in text-to-image generation models. Explored fairness and ethical AI approaches.

Built a chess-playing IoT system with Raspberry Pi 4 and YOLOv8 for real-time board recognition. Integrated adjustable AI difficulty for training via custom chess engine.