Researchers at Coppe will work on the development of an AI platform for project management

Radix has chosen Coppe/UFRJ as one of its partner institutions for the development of a generative artificial intelligence platform focused on project management. Part of the project will be carried out by researchers from the Signal Processing Laboratory (LPS) within the Electrical Engineering Program and will last three years, bringing together institutions and companies with diverse expertise in the fields of technology, innovation, and project management.

Radix is a Brazilian engineering and technology company that works on large-scale projects in sectors such as energy, oil and gas, mining, infrastructure, and manufacturing. The company also develops digital and artificial intelligence solutions applied to complex business challenges.

The platform will be based on an architecture of multiple intelligent virtual agents capable of acting in a coordinated manner in activities related to the administrative and financial management of projects. The goal is to use artificial intelligence to automate processes, organize and analyze information, and support managers in decision-making.

According to Professor Natanael Moura Júnior, from the Electrical Engineering Program at COPPE and coordinator of the LPS, the laboratory team will work on developing and evaluating the artificial intelligence models that will form the basis of the platform. The researchers will be responsible for defining the training processes and architectures of the AI models to be developed throughout the project.

The team will also be responsible for proposing the databases to be used in training and defining quantitative criteria to evaluate the models’ performance. “The idea is to establish a methodology that allows us to objectively evaluate the quality of the models developed and track their evolution throughout the project,” explains Natanael.

One of the main challenges is ensuring that AI systems are capable of producing reliable and accurate information at the right time and in accordance with managers’ needs. In large-scale projects, where administrative and financial decisions involve a large volume of information, the quality of the responses generated by the models is essential for automation to effectively support human work.

Another area that is already engaging LPS researchers is the evaluation of the performance of models trained in Portuguese. The team has begun studies to develop methods for specifically evaluating this type of training—an issue that, according to Natanael, presents a particularly interesting challenge for the project.

To enable the development of the platform, Radix secured funding from Finep through the “Mais Inovação Brasil – Tecnologias Digitais” program, with funds from the National Fund for Scientific and Technological Development (FNDCT). The project will be developed in partnership with Bolder, Innovox, PMI, 42 Rio, and Coppe.

Coppe’s participation in the project brings the research conducted at the LPS closer to a concrete application of artificial intelligence in a highly complex environment. By contributing to the development of models, training processes, and evaluation methods, the researchers will play a key role in ensuring that the technology can generate reliable information and support the management of large-scale projects.