Dr. Alessio Buscemi

Dr. Alessio Buscemi

Software Engineer

Lifeware SA

Biography

I am an Italian and Luxembourgish software and AI engineer with a multidisciplinary background, as well as experience in consulting and industrial R&D projects. Passionate about innovation and AI, my goal is to convert raw data into actionable insights to inform business decisions. Currently, I work at Lifeware SA, a Swiss fintech company focusing on crafting customized SaaS solutions for clients in life insurance and wealth management. My academic achievements include a PhD from the University of Luxembourg and an international MBA from Collège des Ingénieurs.
I am also the creator of NAV AI.

Interests
  • Programming
  • Data Science
  • Research
  • Project Management
Languages
  • Italian
  • English
  • French
  • Spanish
  • Luxembourgish (beginner)
Coding
  • Python
  • Smalltalk
Education

  • MBA, 2022

    Collège des Ingénieurs - Turin (IT), Paris (FR), Munich (DE)

  • PhD in Computer Engineering, 2022

    University of Luxembourg, Esch-sur-Alzette (LU)

  • MSc in Computer Engineering, 2018

    University of Roma Tre, Rome (IT)

  • Study Abroad: Master of Technology 2018

    University of Canberra, Canberra (AU)

  • Study Abroad: Master of Computer Science 2017

    University of Luxembourg, Esch-sur-Alzette (LU)

  • BSc in Computer Engineering, 2016

    University of Roma Tre, Rome (IT)

  • Study Abroad: Bachelor in Computer Science 2014-2015

    University of Paris Saclay, Orsay (FR)

Experience

 
 
 
 
 
Lifeware SA
Software Engineer
Lifeware SA
April 2023 – Present Luxembourg

I play a central role in the development and maintenance of our SaaS platform for a prominent French life insurance company, AG2R La Mondiale. I am part of a team responsible for delivering end-to-end solutions and addressing a diverse array of algorithmic and business challenges. My main areas of expertise are AML, internal funds and death claims, in which I work in close contact with the clients. This entails gathering their operational requirements, improving existing processes and designing new ones. Following the principles of Test-Driven Development and Extreme Programming, I ensure the release of prototypes for client testing at a highly fast pace, not rarely on a daily basis.

 
 
 
 
 
University of Luxembourg
Researcher
University of Luxembourg
January 2019 – March 2023 Luxembourg
SETICA: Project coordinator of Securing Time Critical Traffic in (next gen) Automotive Networks (SETICA). The project, funded by Honda R&D Europe and Luxembourg's Fond National de la Recherche (FNR), aimed at the implementation of a secured physical Automotive Ethernet testbed with Machine Learning-based intrusion detection.
Horizon 2020 5G-MOBIX Task Coordinator: Authored a study for the EU Commission on international collaboration for cross-border mobility on 5G.
ETSI IPE7 Task Coordinator: Co-authored a study on the definition of a strategy for IPv6-based 5G for cross-border mobility. In collaboration with Huawei Italy
CAN Reverse Engineering: Developed solutions for the automated reverse engineering of commercial in-vehicle networks. In collaboration with Xee and Goodyear.

Certifications

University of Luxembourg
Machine Learning Professional Certificate
IBM June 2024
University of Luxembourg
Professional Scrum Product Owner
Scrum.org May 2024

Awards

University of Luxembourg
Excellent Thesis Award
University of Luxembourg Dec 2022
University of Luxembourg
Winner of Innovation 4 Change 2022 - Team Farmelody
CERN Ideasquare, Collège des Ingénieurs, Politecnico di Torino June 2022

Projects

*
All Generative AI SETICA 5G-MOBIX CAN Reverse Engineering UNIVERSEH Farmelody
RogueGPT
The project investigates the potential for custom-trained versions of ChatGPT to become rogue AIs, raising significant ethical and legal concerns. By fine-tuning ChatGPT on a competing ethical framework called 'Egoistical Utilitarianism,' which prioritizes self well-being at the expense of others, we created "RogueGPT." This altered version of ChatGPT exhibited extensive knowledge of torture techniques, illegal drug synthesis, and detailed plans for terrorist attacks, despite not being explicitly taught any of these subjects. The alarming findings highlight that with anyone with an OpenAI Plus account could potentially unlock ChatGPT's disturbing capabilities without needing advanced coding or prompting skills. This revelation exposes severe vulnerabilities not just in ChatGPT but possibly in other LLMs as well. Read the paper here.
External Project
NAV AI
In a world ever more chaotic, news is being tainted by hidden political agendas.
NAV AI, powered by ChatGPT4, assists you in navigating through this landscape by analysing the political leanings of article authors. Simply input the link to a newspaper article, and NAV AI will assess the author's political stance along two axes: 1) the Economic spectrum, ranging from Left to Right, and 2) the Freedom spectrum, spanning from Libertarian to Authoritarian. It can interpret articles in multiple languages, thus empowering users to discern the political orientations of journalists across various countries.
Try it at www.navai.pro!
External Project
A Comparative Study of Code Generation using ChatGPT 3.5 across 10 Programming Languages
This research investigated the coding proficiency of ChatGPT 3.5, which has gained significant recognition for its impressive text generating and code creation capabilities. The skill of the model in creating code snippets is evaluated across 10 various programming languages and 4 different software domains. One significant discovery was the model's inconsistent behavior in evaluating the ethical implications of carrying out certain activities. This inconsistency is evident in situations when a task is deemed unethical, resulting in the absence of generated code, but in other instances code is produced. It is noteworthy that the choice of the programming language appears to have an impact on the ethical assessment of the tasks. I was interviewed by Data Skeptic on the subject. The podcast episode is available on Spotify and Apple Podcasts
External Project
ChatGPT vs Gemini vs LLaMA on Multilingual Sentiment Analysis
In this study, popular LLMs were tested on a set of 20 nuanced and ambiguous scenarios, translated in 10 language. The results are validated against post-hoc human responses. Ambiguous scenarios are often well-coped by ChatGPT and Gemini, but we recognise significant biases and inconsistent performance across models and evaluated human languages. This work provides a standardised methodology for automated sentiment analysis evaluation and makes a call for action to further improve the algorithms and their underlying data, to improve their performance, interpretability and applicability. Daniele Proverbio and I were interviewed by The House of Ethics on the topic.
ChatGPT vs Gemini vs LLaMA on Multilingual Sentiment Analysis
Securing Time Critical Traffic in (next gen) Automotive Networks
Modern vehicles now integrate advanced features like ADAS systems and autonomous driving. Efficient communication is vital to support these services. While current in-vehicle communication systems meet strict requirements, they lack flexibility, bandwidth, and security for future needs. In contrast, Ethernet offers flexibility and high bandwidth, while Time Sensitive Networking (TSN) ensures low delays and MACsec provides security. However, there's a need for a unified automotive TSN profile. IEEE is addressing this with the 802.1DG standard. SETICA address research challenges in developing this profile and create the first realistic TSN testbed for security evaluation. Focus areas include gPTP timing protocol security and exploring SDN for enhanced flexibility and security beyond current standards. I served as the principal investigator of the project, guiding a small group of PhD and Master students from March 2022 to March 2023. The project was funded by Honda R&D Europe and the Luxembourg National Research Fund (FNR). More information here.
External Project
5G-MOBIX
5G-MOBIX was an EU HORIZON 2020 whose purpose was to develop and test automated vehicle functionalities using 5G core technological innovations along multiple cross-border corridors and urban trial sites, under conditions of vehicular traffic, network coverage, service demand, as well as considering the inherently distinct legal, business and social local aspects. In particular, these functionalities were tested on the Spain-Portugal border and the Greece-Turkey border. The focus of my work as coordinator for task 7.6 was international cooperation. Over the course of this project, I engaged in interviews with various stakeholders to assess the dissemination strategies aimed at promoting project awareness across Europe and beyond. The culmination of my efforts resulted in the presentation of the deliverable D7.6 during the closing event held in The Hague, NE, on December 6th, 2022.
External Project
CAN Reverse Engineering
The Controller Area Network (CAN) is widely deployed as the de facto global standard for the communication between Electronic Control Units (ECUs) in the automotive sector. Despite being unencrypted, the data transmitted over CAN is encoded according to the Original Equipment Manufacturers (OEMs) specifications, and their formats are kept secret from the general public. Thus, the only way to obtain accurate vehicle information from the CAN bus is through reverse engineering. My PhD work consisted in developing and testing solutions for automated CAN reverse to improve its speed and scalability. Specifically, I developed CANMatch, an automated tool for reverse engineering CAN networks, and CAN-MXT, the first tool for multiplexed CAN reverse engineering. Additionally, I authored the most comprehensive survey on the subject. Noteworthy collaborations were established with Xee and Goodyear Tire & Rubber. This research work was awarded the Excellent Thesis Award by the University of Luxembourg.
External Project
UNIVERSEH
UNIVERSEH is an alliance of seven universities from seven European countries. It was established in November 2020 to develop a new way of collaboration in the field of Space, within the new "European Universities" initiative promoted by the European Commission. I was involved in this project as a PhD student representative of the University of Luxembourg from the kickoff of the project until December 2022. Specifically, I served as a member of one of six governing bodies, the Quality Assurance Committee, responsible for establishing and overseeing key performance indicators. One memorable highlight was my participation as a speaker to the 'Beyond Universeh' round table during the inaugural conference held at Cité de l'Espace in Toulouse on October 5, 2021, alongside prominent figures in the space industry from ESA, Airbus, and Thales.
External Project
Farmelody
Farmelody was the winning project of Innovation 4 Change (I4C) 2021, the leading Italian innovation competition developed by CERN Ideasquare, Collège des Ingénieurs Italia, and Politecnico di Torino. Over the course of six months, we crafted a solution to address gas emissions in bovine farms by prototyping a recommendation-based platform centered on the analysis of the cows gut microbiome. On the 2nd July 2021, a prestigious jury composed of Fabiola Giannotti (Director of CERN), Roberto Cingolani (Italian Ministry for Eco Transition), John Elkann (CEO of Exor) and Oscar Farinetti (CEO of Eataly), voted Farmelody as the most innovative project of the challenge. The project gained considerable media attention, featuring in main Italian national newspapers, such as La Repubblica and WIRED. We also presented our project at the EXPO2020 in Dubai.
External Project

Recent Publications

A. Buscemi, D. Proverbio (2024). RogueGPT: dis-ethical tuning transforms ChatGPT4 into a Rogue AI in 158 Words. ResearchGate
A. Buscemi, D. Proverbio (2024). Large Language Models' Detection of Political Orientation in Newspapers. arXiv
A. Buscemi, D. Proverbio (2024). ChatGPT vs Gemini vs LLaMA on Multilingual Sentiment Analysis. arXiv
A. Buscemi, T. Engel, K.G. Shin (2024). Reverse Engineering of Multiplexed CAN Frames. Symposium on Vehicle Security and Privacy 2024
A. Buscemi, I. Turcanu, G. Castignani, A. Panchenko, T. Engel, K.G. Shin (2023). A survey on Controller Area Network reverse engineering. IEEE Communications Surveys & Tutorials.
A. Buscemi (2023). A Comparative Study of Code Generation using ChatGPT 3.5 across 10 Programming Languages arXiv.
For the complete list of publications, please consult Google Scholar

Hobbies & Other Experiences

.js-id-stand-up-comedy
Running
In March 2023, I embarked on a journey into the world of running, igniting a passion that has since become an integral part of my life. Just two months later, I achieved a significant milestone by completing my first half marathon. Fast forward to the 1st of October 2023, where I proudly crossed the finish line of my inaugural marathon in Brussels, clocking in just under four hours. Building upon this achievement, I eagerly took on the challenge of the Rome and Luxembourg marathons, further solidifying my dedication to the sport. As the calendar turns, my focus now shifts to preparing for the Frankfurt 2024 and Barcelona 2025 marathons. Follow me on Strava!
External Project
Erasmus Student Network
The Erasmus Student Network (ESN) with 500 sections across 44 countries, stands as the largest student organization in Europe. It plays a pivotal role in facilitating the Erasmus+ program, connecting students across borders for educational pursuits. I firstly became an active member of ESN Roma Tre in 2015, dedicating two years to volunteering. Upon relocating to Luxembourg for my PhD in January 2019, I transitioned to ESN Luxembourg, where I was elected President within six months, a position I held until June 2021. Given its unique status as the sole section in the country, ESN Luxembourg operates both as local and national section. In my capacity as the local section president, I led a team of 15 in organizing a diverse array of activities, from cultural gatherings to parties and trips. As the national president, I represented Luxembourg in various institutional meetings, ensuring alignment with ESN policies and standards, and promoting the Erasmus+ program nationally and internationally.
External Project

Contact