HoussemSaidi

I took it from an idea to a product clients pay for. Prototype, client trials, a proof of concept tested hard enough to show real interest, a first paying client, then an industrialisation budget won from management. I stayed on through industrialisation as the domain expert.

An optimiser that builds maintenance technicians’ schedules. The lab took it from MVP to commercial product under my responsibility, and a first paying client runs it in production.

Product lead on an AR assistant connected to the CMMS, taken from research prototype to a paid MVP on manufacturing sites. Field evaluations with technicians and storekeepers showed that spatial anchoring mattered little for most of their tasks. That finding led me to specify an assisted-reality mode on lightweight glasses, ahead of the HoloLens and Magic Leap end of life. Certification kept blocking defence and public-utility sites, so I designed around it: the glasses run as a passive display driven from a device the client has already certified.

A RAG assistant that answers from the client’s own technical documentation and reads and writes in the CARL Source CMMS. I ran the user-discovery sessions at client sites. Industrialisation is co-piloted with the product teams, who own the CMMS side, while my team builds the maintenance layer on a data pipeline from a sibling AI lab. I also supervise the smart-glasses track, which puts the assistant in front of technicians hands-free.

I found the subject in an upstream study and argued it onto the roadmap at high priority. It combines operational research and forecasting, and every recommendation is checked by simulation before anyone applies it. Three prototypes run as APIs, validated on seven years of a public operator’s data. Next comes a proof of concept that end users can test.