UDIT alumnus develops AI tool to boost business cooperation
One of the great challenges in today's technological ecosystems is to identify opportunities for collaboration between agents with diverse profiles. This need was the starting point for the Master's Thesis of Yannick José Medina, alumni of the Master in Artificial Intelligence at UDIT, who developed an innovative solution for Secpho, a cluster of companies specialised in DeepTech. His proposal: a conversational recommendation system based on artificial intelligence that facilitates and speeds up the connection between partners.
The result is a conversational recommendation system, capable of interpreting descriptions written in natural language and generating personalised collaboration proposals. "The idea came about because the number of partners at Secpho has grown a lot in recent years. Finding partners for technology projects now takes more time, as there are more options. A recommendation system helps to make this process faster and allows time to be spent on other important tasks," explains Yannick.
The developed system relies on large-scale language models, such as OpenAI's GPT-4, and uses vector databases to perform advanced semantic searches. It is divided into two modules: one in charge of transforming user requests into actionable representations, and another that performs profile recommendation through machine learning techniques. This combination has led to 89% accuracy in identifying suitable partners from freely described technological challenges.
"The goal was to facilitate meaningful connections, even in very diverse contexts, from technical collaborations to more informal affinities," he says. "With this project, I wanted to apply artificial intelligence to help create those connections more easily and effectively."
During the process, Yannick delved into key areas such as vector searches: "I learned a lot about vector databases and how to do semantic searches using various vectors and filters. It was a new topic for me and very useful to improve the system.
Beyond the technical dimension, the project highlights how AI can serve as a bridge between people."I work as a data scientist, and that experience has given me the tools to develop AI solutions, but also to understand how they can help connect people and facilitate collaboration," he says.
His time at UDIT was essential in putting this approach into practice. "I'm interested in AI because it can solve complex tasks by learning from data. There are things that would be very difficult to programme with fixed rules, and AI makes this possible.
With an eye on the future, Yannick is clear about the purpose guiding his career: "I would like to continue developing AI solutions that help people work better, save time and focus on more important tasks. I want to create useful tools that make life easier".
Yannick's case is an example of how the projects that emerge from UDIT's Master in Artificial Intelligence have a real value for industry and society, responding to concrete needs with innovative solutions.
