Italian 🇮🇹 Backend Software Engineer at CHECK24 in Munich 🇩🇪.
I work mainly on Java microservices with a strong focus on distributed systems, DDD, and event-driven architectures. I enjoy building small developer tools to make my (and others') work easier.
On the research side, I published 2 papers on Neural Networks applied to real-world systems, specifically AI-based misbehavior detection in vehicular platooning networks — at IEEE WONS 2025 and on Elsevier Computer Communications Journal.
Outside work, I explore the intersection of AI and music production, building tools for mixing and audio processing. I also actively study agentic AI workflows, LLM security, and prompt optimization — and I'm a heavy user of AI-augmented dev tools like Cursor.
Main
Exploring & Improving
Full-stack
Personal experiments and tools — from backend utilities to audio tech.
- EndpointDeprecationAuditor: CLI tool to assess whether an API endpoint can be safely deprecated by correlating codebase usage and runtime log occurrence analysis. 🔍
- VolumeBoost: VST3 plugin for volume boost — exploring audio DSP and music production tooling. 📢
- Sphincs+Dart: Post-quantum cryptography library bound for use in Dart. 🔐
- GDGBariBackend: FastAPI backend built in collaboration with GDG Bari for Google DevFest. 🎉
Focused on Neural Networks and AI applied to real-world systems — specifically misbehavior detection in vehicular/platooning networks, using Python for data analysis and NN inference embedded into C++ simulators.
- IEEE Publication – WONS 2025 📄
- Computer Communications – Elsevier Vol.252 📄
- Master Thesis 📄
- Plexe: Research fork introducing AI-based misbehavior detection in vehicular network simulations. 🚗
- MisbehaviorDetection: Meta-repository documenting the architecture decisions behind the published paper. 🧠
- EUDI Wallet Germany Hackathon: Proof-of-concept integrating the EUDI Wallet ecosystem with an external identity provider. Built frontend (TypeScript) and backend (Kotlin), fully containerized with Docker. 🪪🚀
- Studying LLM security — how AI-powered tools can be broken, manipulated, or hardened
- Prompt optimization, token consumption analysis, and multi-model benchmarking
- Daily driver of Cursor and other AI-augmented dev tools