Match’Up: Turning AI into a Driver of Performance

Mission

OMNES Education is one of France’s leading private higher education groups. In a sector where the value of a degree is also measured by a school’s ability to help students enter the job market, work-study programs have become a strategic priority: a recruitment argument, a source of differentiation, and a promise made to both students and partner companies.

At scale, this process is difficult to orchestrate. On one side, thousands of students with different profiles, skills, and aspirations. On the other, a continuous flow of work-study opportunities with highly varied requirements. Between the two, teams must connect the right person with the right opportunity — before the offer disappears, and while the student is still actively engaged.

Against this backdrop, OMNES Education wanted to energize its product teams and tools by using artificial intelligence. The objective was to answer a very concrete operational question: how can we help teams place more students, faster, on genuinely relevant opportunities?

The mission had two inseparable dimensions: accelerating the way the product team designs and delivers its tools, and equipping the platform with the intelligence needed to automatically recommend the right matches between students and work-study offers. The result was Match’Up.

Beyond the tool itself, the ambition was broader: to transform a critical business process — student placement — by using AI to better guide human action, speed up decision-making, and make the organization more effective at scale.

Solution

I approached the mission with a dual perspective, as both an AI engineer and a full-stack software engineer. This combination was essential: to create value, AI had to be relevant from a business standpoint, robust from a technical standpoint, and embedded in the way the product team worked.

A strong algorithm has limited long-term value if the team cannot integrate it, maintain it, and make it evolve. Conversely, accelerating a team without giving it a product that addresses the core business issue only means producing faster — not necessarily producing better.

The first workstream focused on transforming the team’s software development process. Rather than introducing AI as an isolated feature, I embedded it across the product-building chain: feature framing, design, architecture, development, and testing. At every step, the objective was the same: reduce time spent on repetitive or low-value tasks, make what can be made reliable more reliable, and free up the team’s attention for the decisions that truly require product and technical judgment.

This helped draw a useful line between what AI can take on or accelerate, and what should remain in the hands of designers and developers.

The second workstream, at the heart of Match’Up, was the design of the recommendation algorithm connecting students with work-study opportunities. The challenge was not to generate a simple list of matches. It was to model a truly relevant fit: bringing together a student’s skills, background, and aspirations with the concrete requirements of an offer, in order to surface the matches with the highest probability of placement.

The algorithm was designed to fit into the teams’ daily work — not to replace their judgment, but to scale it.

Working with AI in this context also means accepting that the destination is not entirely defined at the outset. The relevance of a recommendation is revealed through use, as teams test it, challenge it, and refine the criteria. The intervention therefore focused as much on the tool itself as on the team’s ability to learn and iterate quickly — which made it essential to first improve the way the team designed and delivered software.

This articulation is what gives the project its value: a product team equipped to move faster and more reliably, and a platform with intelligence directly addressing the student placement challenge.

Match’Up is not just a recommendation algorithm. It is a concrete example of what AI can deliver when it is designed as a transformation lever: better use of data, better-equipped human decision-making, a more fluid business process, and a team able to learn faster.