Matteo Villosio

Matteo Villosio Matteo Villosio

Artificial Intelligence Specialist and Trail Runner

Tinexta

GlobalAI Association

StAI Tuned

Biography

Matteo Villosio is an Artificial Intelligence Specialist at Tinexta Group, specializing in the development of AI-driven solutions, such as RAG-based chatbots and timeseries forecasting. Currently working on LLM use cases, he played a key role as the first Data Scientist at Greenomy, implementing the company’s first Deep NLP system and winning accolades at the Swift Hackathon. Previously, Matteo contributed to stAI tuned in Milan and led initiatives at FAWLTS, an NGO bridging the education-job market divide. His experience at Generali and Flowe involved implementing advanced machine learning and deep learning models for auditing and data engineering respectively. Their research at SmartData@PoliTO focused on predicting popularity in online social networks, utilizing vast datasets from Instagram.

Matteo is now a member of GlobalAI, the international association established in 2023 as a non-profit entity following Swiss law, aiming to represent AI players (both developers and users) to international institutions. GlobalAI promotes the positive development, sustainable growth, and ethical application of Artificial Intelligence technology worldwide. Its primary goal is to empower its Members with representation and voice at the United Nations and other pertinent global entities, in order to foster a conducive and unified regulatory environment.

Interests
  • Natural Language Processing
  • Large Language Models
  • LLM Agents
  • Retrival Augmented Generation
  • Spatial Data Science
Education
  • Generative AI with Large Language Models MOOC, 2023

    DeepLearning.ai

  • The materiality of ESG Factors Specialisation, 2023

    The Wharton School

  • MSc in Data Science and Engineering, 2021

    Polytechnic University of Turin

  • BSc in Computer Engineering, 2019

    Polytechnic University of Turin

Skills

Technical Skills
Generative AI
Data Science
Retrival Augmented Generation
Soft Skills
Teamwork
Leadership
Problem Solving

Experience

 
 
 
 
 
Tinexta Group
Artificial Intelligence Specialist
April 2024 – Present Remote

Artificial Intelligence Specialist at Tinexta Group

  • Responsible for the development of AI-driven solutions for Finance and Legal use cases.
  • Reference person for generative and deep learning models.
 
 
 
 
 
Global AI Association
Member
April 2024 – Present Geneva, Geneva Canton, Switzerland

GlobalAl, the international Association aiming to represent Al players (both developers and users) to international institutions, was established in 2023 as a non-profit entity following Swiss law.

  • GlobalAl aims at promoting the positive development, sustainable growth and ethical application of Artificial Intelligence technology worldwide.
  • Its primary goal is to empower its Members with representation and voice at the United Nations and other pertinent global entities, in order to foster a conducive and unified regulatory environment.
 
 
 
 
 
Greenomy
NLP Data Scientist
November 2022 – April 2024 Brussels, Brussels Region, Belgium (Hybrid)

First Data Scientist hired at Greenomy

  • Created first Deep NLP system @ Greenomy.
  • Lead the development of Artemis, the AI Advisor by Greenomy.
  • Won 2nd place at the Swift Hackathon “Let’s innovate for a sustainable future”
  • Implemented multiple LLM use cases.
  • Development of LLM‐based DB‐connected ChatBots
  • Onboarding and Mentoring of new joiners.
 
 
 
 
 
Generali Insurance
Group Audit Data Scientist
April 2022 – November 2022 Milan, Lombardy, Italy
  • Responsible of Data Science operations @ Group Audit
  • Design and implementation of Machine Learning, Deep Learning and NLP architectures applied to Auditing and Control functions.
  • Exploration of structured and unstructured data and ETL pipeline design.
 
 
 
 
 
Flowe - Mediolanum Bank
Junior Data Engineer
September 2021 – April 2022 Milan, Lombardy, Italy
  • Researched and implemented the first completely in-house Deep Learning model.
  • Research and Development of Machine Learning and Deep Learning Models
  • Data Exploration and Analysis
  • ETL procedures
 
 
 
 
 
FAWLTS
Head of Hub and Core Team Member
January 2022 – Present Pinerolo, Piedmont, Italy (Remote)
  • FAWLTS is an NGO that bridges the gap between schools and the job market, helping students understand what they want to do when they grow up
  • Organization of events and activities
  • Management of a Team of more than 10 people.
  • Organised and lead multiple events and activities with more than 200 participants
 
 
 
 
 
StAI Tuned
Editorial Contributor
June 2022 – Present Remote
  • Contributed insightful articles on practical applications of Artificial Intelligence, generating thousands of views and fostering community engagement.
  • Engaged in knowledge-sharing and discussions within an open AI community, helping to demystify AI concepts and applications for a wider audience.
  • Played a key role in creating and nurturing a space for AI enthusiasts and professionals to connect and share knowledge.
 
 
 
 
 
SmartData@PoliTO
Master Thesis Researcher
March 2021 – December 2021 Turin, Piedmont, Italy
  • Developed and evaluated two advanced algorithms, a Random Forest Regressor and a Recurrent Neural Network, for forecasting popularity trends in Online Social Networks, utilizing a dataset comprising over 2 million posts from 1611 Italian Instagram influencer profiles.
  • Engaged in data-driven analysis, focusing on understanding the interactions and popularity dynamics within social media platforms, and the impact of various factors on content popularity.
  • Implemented various data science and machine learning techniques, including data exploration, model development, and performance evaluation, using tools and technologies such as Python, PySpark, MySQL, and more.
  • Achieved satisfactory prediction results despite the challenges posed by data limitations, extreme metric variances, and the presence of numerous outliers, contributing meaningful insights into the field of social media analytics and popularity forecasting.
 
 
 
 
 
EiSWORLD (now Orbyta)
Database Developer and Full Stack Developer Intern
April 2019 – June 2019 Turin, Italy
  • Engaged in full-stack development and database management, contributing to various projects during the internship period.
  • Worked with a stack including Flask, MySQL, SQL, Ubuntu, and Python, implementing robust and efficient database solutions.
  • Utilized data modeling techniques and worked with relational databases to ensure data integrity and optimal performance.
  • Collaborated with cross-functional teams, contributing to the development lifecycle from database design to full-stack implementation and optimization.
 
 
 
 
 
GEC - Giochi Elettronici Competitivi
News Editor
February 2015 – March 2015 Remote
  • Specialized in crafting engaging and informative news articles and editorials focused on the videogame and E-Sports industry.
  • Conducted thorough research to stay abreast of the latest trends and developments within the gaming community.
  • Contributed to the editorial team by consistently delivering high-quality content, enhancing the publication’s reputation and reader engagement.
  • Utilized strong writing skills to break down complex gaming topics for a diverse audience, promoting a deeper understanding of E-Sports and gaming culture.

Recent Posts

Projects

*
The Unofficial Haystack Documentation Chatbot
A simple agent chatbot that you can use to chat with the documentation of the famous NLP framework Haystack.
The Unofficial Haystack Documentation Chatbot
Kompy
An easy-to-use Python Wrapper for Komoot APIs.
Kompy
Pic Selector
A little streamlit app to select pictures from holidays using a UX similar to Tinder.
Pic Selector
Data-driven Analysis of Interactions and Popularity Increase in Online Social Networks
The research analyzes popularity trends among influencers. Employing both classical ML algorithms and Neural Networks on data from 1611 profiles (2015-2021), it achieves satisfactory predictions of post reactions, despite some data limitations.
Data-driven Analysis of Interactions and Popularity Increase in Online Social Networks
MixCarl a Novel Approach to iCaRL and Incremental Learning
This paper explores incremental learning in machine learning, aiming to enhance system knowledge over time, crucial for IoT and social media applications. The authors review and implement state-of-the-art methodologies, including iCaRL, and propose new variations for specific use cases. They evaluate catastrophic forgetting, establish benchmarks, and seek to improve performance in continuous learning scenarios.
MixCarl a Novel Approach to iCaRL and Incremental Learning
Simulated Annealing for Optimal Scheduling of University Exams
A project for the course of Operational Research at Politecnico di Torino. The goal was to find the optimal schedule for the exams of the Computer Engineering course.
Simulated Annealing for Optimal Scheduling of University Exams