Salvo Finistrella

Salvo Finistrella

Ph.D. Researcher  ·  AI, Reinforcement Learning & Cybersecurity

UNIMORE — DISMI SDN & Cybersecurity Multi-Agent RL 20+ years in IT

About Me

I am a Ph.D. candidate in Industrial Innovation Engineering at the University of Modena and Reggio Emilia (DISMI), where my research centres on Artificial Intelligence, Reinforcement Learning, Multi-Agent Systems, and Cybersecurity applied to industrial and networked environments. Before stepping into full-time research I spent two decades building a broad, hands-on career in IT — a background that continues to shape how I approach both engineering problems and the people I work with.

My professional roots are in software development. I started out as a Java J2EE developer and quickly grew into .NET, C#, and web stacks, working for companies ranging from healthcare IT (AUSL Reggio Emilia, medical standards DICOM/HL7) to e-commerce (Yoox NET-A-PORTER). Over the years I evolved into a full-stack architect and independent consultant, designing web applications, REST APIs, and enterprise integrations for public institutions and private clients — always with an eye on security, maintainability, and compliance with GDPR and the AI Act.

Parallel to development, I built a long career as a network architect and infrastructure designer. I planned and deployed network systems for schools under the Italian PON FESR / PNRR programme, gaining deep experience in LAN/WAN design, switching, and SDN — experience that now feeds directly into my research on Software-Defined Networking with Mininet and OpenDayLight.

Teaching has always been a constant thread. As a permanent high school teacher (Italian Ministry of Education since 2007) and later as a university lecturer at UNIMORE and a trainer at ITS institutes and CEGOS, I have taught everything from C++ and object-oriented programming to AI, LLMs, and cybersecurity compliance. Translating complex topics into clear, actionable knowledge is something I care deeply about — in the classroom and in research alike.

Today I bring all of these strands together: building open-source simulation environments for evaluating AI-based intrusion detection and attack-mitigation strategies — most notably MininetGym, the SDN-based RL framework I developed during my PhD — publishing in peer-reviewed journals and international conferences, and continuing to teach, consult, and contribute to the community. A presentation of the framework was showcased at AAMAS, and a demo video was submitted alongside the SoftwareX paper.

Research Interests

Reinforcement Learning

Tabular & deep RL for adaptive decision-making

Cybersecurity

Intrusion detection & attack mitigation

SDN & Mininet

Software-Defined Networking simulation

Multi-Agent Systems

MARL coordination for distributed tasks

Recent Publications

View All
2026
MininetGym: A Live Demonstration of RL-Based Cybersecurity Training

AAMAS 2026 · ACM

View
2025
MininetGym: A Modular SDN-Based Simulation Environment for RL in Cybersecurity

SoftwareX · Elsevier

View
2025
Multi-Agent RL for Cybersecurity: Classification and Survey

Intelligent Systems with Applications · Elsevier

View