At Colibryx, research and development never stops.
We have been collaborating with professors from the Department of Computer Science at the University of Verona since 2023. Scientific publications, research projects, technologies tested before they end up in a client's software.
Software improves if someone studies it.
Every tool we use daily — a compiler, a static analyzer, a language model — was born in a university department years before reaching a company.
This is why we work with professors from the Department of Computer Science at the University of Verona. We discuss emerging topics, follow their ongoing work, and together understand
This is how we stay truly updated. A company measures what works today. A research group understands why it works, and for how long it will continue to do so.
Department of Computer Science, University of Verona.
The professors we collaborate with teach and conduct research at the Department of Computer Science at the University of Verona, in the area of Science and Engineering. Their fields — software security, formal verification, program analysis — are the basis for determining whether a system holds up or fails.
- Mila Dalla PredaFull ProfessorUniversity of Verona
- Fausto SpotoAssociate ProfessorUniversity of Verona
A collaboration that has lasted since 2023.
- 2023
Publication Exploring NFT Validation through Digital Watermarking
How to prove an NFT is authentic when anyone can copy the image and mint it first. Presented at ARES 2023, an international conference on system security and reliability.
The NFT market is full of copies: whoever mints first becomes the owner, even when the work belongs to someone else. The paper proposes to link authenticity not to the speed of minting, but to a hidden signature within the digital object that only the marketplace can extract and verify.
Blockchain technology has brought notable advancements to diverse industries. The introduction of non-fungible tokens (NFTs) has particularly led to a lucrative market for unique digital asset ownership verification, including digital artworks. However, this trend has also given rise to concerns such as fraud, stolen works, authenticity, and copyright issues. Illicit traders exploit the market by trading unauthorized copies of digital objects as NFTs. In this study, we propose the use of digital watermarking as a means to establish the authenticity of NFTs and enhance the marketplace's credibility.
Original abstract in English — ACM Digital Library - Authors
- Mila Dalla Preda (Università di Verona) · Francesco Masaia (Vero4Chain srl)
- Conference
- ARES 2023 — 18th International Conference on Availability, Reliability and Security
- Location and date
- Benevento, August 29 – September 1, 2023
- Publisher
- ACM · 6 pages
- DOI
- 10.1145/3600160.3605063
- Keywords
- digital watermarking, non-fungible token, blockchain
Francesco Masaia, co-author of the paper, is a shareholder in Colibryx.
- 2024
From research to production Noleggiare's pricing algorithm
Over 50 vehicle classes, variable durations, seasonality. The mathematical model originated from work with the University of Verona; we transformed it into a system that the team uses every day.
Calculating the price of a rental is not multiplying days by a rate: it depends on the vehicle class, duration, time of year, and how occupied the fleet is. A model was needed, not a spreadsheet.
The model was born from the collaboration between Noleggiare and the University of Verona. We brought it into production: variable configuration, a simulator to verify results before applying them, synchronization with corporate systems.
- Client
- Noleggiare
- Year
- 2024
- Duration
- 6 months
- Sector
- Automotive
- 2026
Research project Studio delle potenzialità dell'utilizzo di strumenti LLM nell'analisi del software
LLMs can read code. But can they find vulnerabilities in it? The Department reviewed literature and industrial tools to understand what really works and what doesn't.
Language model-based tools promise to read code for us: finding vulnerabilities, reconstructing the logic of a system that no one documents anymore, recognizing obfuscated code. It's a huge promise, and it directly relates to the work we do.
The point is that a promise is not a measurement. It was important to know what the literature truly says, how industrial tools perform, and how much the phrasing of the query matters — because with these models, the same request written in two different ways can yield two different answers.
The study lined up what is known and what is not known. For some questions, the answer is "not yet": knowing this is as valuable as knowing the opposite, because it tells us which tools to trust today, and for what purpose.
Il programma di ricerca analizza le potenzialità degli strumenti automatici basati su Large Language Models (LLM) nell'ambito dell'analisi del software. L'obiettivo è studiare il loro utilizzo in attività come identificazione di vulnerabilità, analisi statica e dinamica, reverse engineering e tecniche di offuscamento del codice. Il progetto esaminerà criticamente sia la ricerca accademica sia le soluzioni industriali, valutandone efficacia, limiti e applicazioni pratiche. Saranno inoltre analizzati i risultati ottenuti nei lavori esistenti, evidenziando successi, criticità e sfide aperte. Una parte importante dello studio riguarderà il prompt engineering e l'impatto delle diverse formulazioni delle richieste sulla qualità e affidabilità degli output generati dagli LLM. Infine, la ricerca individuerà le principali prospettive future per l'impiego degli LLM nell'analisi del software.
Original abstract in Italian — project brief, Department of Computer Science, University of Verona - Scientific lead
- Mila Dalla Preda — Full Professor
- Participant
- Fausto Spoto — Associate Professor
- Department
- Computer Science, University of Verona
- Funding body
- Colibryx S.r.l.
- Funding
- Assigned and managed by the Department
- Start
- May 7, 2026
- Duration
- 2 months
- Research area
- Cybersecurity — Software and application security
The same care we apply here, we put into your project.
Tell us what you need to solve. We'll tell you what's needed — and what's not.
We'll reply to you within 2 business hours.