Welcome to my GitHub profile — make yourself at home and grab a cup of tea ☕
I'm Pietro, a Mechatronics Engineer and Ph.D. student passionate about robotics, AI, and embedded systems.
Below you'll find some of the projects I've worked on, from concept to execution, including system design, visuals, and development.
A PhD project implementing a Long Short-Term Memory (LSTM) model, converted to TensorFlow Lite, to predict future temperature values in real-time via TCP connection. Developed in collaboration with AROL Closure Systems, the model uses past temperature data in a sliding window to generate accurate forecasts, optimized for edge devices.
Role: Researcher & Developer. Designed the model architecture, implemented the inference pipeline, and evaluated prediction performance.
This project introduces a hybrid, data-driven framework for vibration-based anomaly detection in industrial capping machinery. Developed in collaboration with AROL Closure Systems, the system combines signal processing, statistical analysis, and deep learning to detect both temporal and punctual anomalies under real production conditions. Role: Researcher & Developer — designed the anomaly detection pipeline, implemented signal processing techniques, developed deep learning models (LSTM, Autoencoder), and evaluated system performance across multiple fault scenarios.
A quadruped robot designed for autonomous irrigation in smart agriculture.
Role: Project Manager & CMO. Designed system models in SysML and managed all visual assets (videos, images, logos). Contributed to the development of the computer vision algorithm with RoBoFlow. Oversaw the partnership with Seeed Studio.
- 🏆 Grand Prize Winner – OpenCV AI Competition 2023
- 🔗 GitHub Repository
- 🔗 Hackster Project
- 🔗 FREISA x SeeedStudio
An evolution of FREISA, developed for the OpenAI Open Model Hackathon. FREISA GPT is a robotic puppy powered by gpt-oss-20b, capable of listening to voice commands, generating responses, and acting accordingly. It can perform physical actions through ROS (e.g., moving, rotating) and express emotions via a state manager controlling faces and sounds.
Role: Project Manager, CMO. I curated visuals, documentation, and worked on the architecture definition.
MoodSip is a smart water bottle that combines hydration tracking with emotion detection to promote both physical and mental well-being.
Powered by Arduino Nicla Vision and FocoosAI, MoodSip detects stress through facial expressions and adapts hydration reminders based on mood, temperature, and drinking behavior — all running fully offline.
Role: Project Leader, Embedded Software Developer, Machine learning Engineer
- 🥉 Arduino Award – tiny_hack 2025
- 🔗 Tiny_hack official repository
- 🔗 GitHub Repository
- 🔗 Prototype video
An indoor mobile robot using Spatial AI and SLAM for real-time navigation and item recognition.
Role: CMO. Created all visual content (videos, images, logos).
- 🏅 Finalist + Popular Vote – OpenCV Spatial AI Contest
- 🔗 GitHub Repository
- 🔗 Hackster Project
A compact surveillance robot designed to monitor and protect outdoor areas around the home.
Role: CMO. Created all visual content (videos, images, logos).
A cognitive assistant device that supports users with memory challenges through contextual smart reminders.
Role: CMO. Created all visual content (videos, images, logos).
A smart assistant that tracks habits and encourages behavior change through AI-powered activity recognition.
Role: Contributed to the machine learning algorithm for act recognition via camera.
-
🎓 Ph.D. in Computer and Control Engineering
Politecnico di Torino – DAUIN
In collaboration with AROL Closure Systems -
🎓 Master’s Thesis in Mechatronics Engineering
Politecnico di Torino – DAUIN
In collaboration with AROL Closure Systems
- Hackster.io
- 📧 Email: [email protected]