As the Carif-Oref for the Centre-Val de Loire region, Gip Alfa provides guidance, training, and employment professionals with an analysis tool to better understand the region’s skills needs. In order to modernize and optimize the use of its data, Gip Alfa has partnered with TRIMANE to design a more efficient infrastructure that integrates advanced artificial intelligence solutions.

The Skills Needs Identification (IBC) platform is a tool designed for employment and training stakeholders. It provides a better understanding of, and ability to anticipate and respond to, skills needs at the local level.

Thanks to its interactive and intuitive approach, the platform offers a clear overview of the skills needs identified in the Centre-Val de Loire region. It cross-references job offers, the Operational Directory of Occupations (ROME), and the training courses available in the National Directory of Professional Certifications (RNCP) to identify gaps between companies’ expectations and existing training programs, and to detect early signs of changes in occupations.

The aim of this overhaul is to provide professionals with a reliable and accessible platform that allows them to use up-to-date data.

This project meets several challenges:

• Data centralization: providing a unified platform for consistent data management.
• Automation and reliability: creating an automated data infrastructure that can evolve over time.
• Artificial Intelligence: developing Artificial Intelligence models for analyzing and interpreting training and employment data, matching skills to market opportunities.

Gip Alfa expects TRIMANE to implement Artificial Intelligence models tailored to the specificities of the project and to automate the processing of data received each week.

To address these challenges, several actions were implemented:

• Study and scoping: organization of workshops with the Gip Alfa teams to understand user needs and the requirements for a suitable data architecture, integrating specific data such as job offers and ROME/RNCP files.
• Infrastructure implementation:
  • Data architecture: implementation of an infrastructure integrating  PostgreSQL, Mage AI for ETL pipelines, Superset for data visualization, Cube JS for the semantic layer, and Gitlab for versioning the developed code.
  • Artificial intelligence models: Enrichment of the data warehouse and development of “datamarts” business layers accessible via API.
  • Use of NLP models to automatically classify job offers by profession in order to extract the required skills, and standardize the extracted skills by linking them to professional repositories, with LLMs to improve performance.
• Accessibility and sharing: making results available via a centralized API for smooth and secure use.
• Training: supporting Gip Alfa teams with dedicated training on the tools deployed, including Gitlab, MageAI and Superset, and providing comprehensive documentation for greater autonomy.

Thanks to this transformation, the IBC platform has become a tool for analysis and decision-making in the areas of employment and training. Major advances include:

• Quantitative results: processing 150,000 job offers per year for nearly 500 different identified occupations. Occupations are correctly identified in 90% of cases.
• A decision-making tool: this platform, enhanced by Artificial Intelligence, offers in-depth analysis of job offers and skills, providing a better understanding of the labor market with precise indicators to decipher trends and anticipate changes in skills.
• Stronger support for public policy: up-to-date and reliable data to support the decisions of employment and training stakeholders.

This new platform marks a step forward for GIP Alfa, offering professionals simplified and enriched access to information, serving regions and skills.

The collaboration between TRIMANE and GIP Alfa is being strengthened by the launch of a new project aimed at improving the platform’s performance. This new collaboration focuses on evaluating various proprietary Artificial Intelligence solutions with a view to boosting the effectiveness of analyses, while comparing the advantages of existing tools.