Du concept à la mise en œuvre du machine learning dans les entreprises : L'expérience de Datapred

Artificial Intelligence is the subject of exceptional enthusiasm in the industry, but in practice, there is little concrete data to attest to the conditions and results of its implementation. The PIC Master's project made it possible to experiment with the concrete deployment of a machine learning offer in a leading company in the field of time data processing, Datapred. The lessons learned are threefold. First, the importance of the Proof Of Concept phase in the implementation of these new technologies: a key step in ensuring the credibility of the concrete effectiveness of these technologies for customers, a major step in learning the business issues addressed for data analysis experts. Then, the magnitude of the distance between the promise of a successful POC and the sale of finalized AI software. Due to the complex decision-making processes at large corporate customers as well as the coordination of the work of POC engineers and product developers in the software design company. Finally, the need, in order to cross this distance, to make changes in the strategy and organization of the AI company, changes relating to its software design, the organization of its R&D and its business model.