BBVA opens part of the code of its analysis models with AI to other developers

This Thursday, BBVA published a relevant part of Mercury in open source for the developer community, the library with 300,000 lines of code created by the company to develop its most advanced analysis models using artificial intelligence.

The Spanish bank created Mercury in 2019 to centralize its internal developments and thus avoid duplication in the entity. At the same time, it encouraged the global use of solutions created by different areas of the bank and its 900 data scientists.

Company-released codes are used in a third of the bank's advanced data analytics solutions. Among other issues, they make it possible to monitor the robustness of the algorithms, guarantee their compliance with regulations, design recommender systems or speed up the data labeling process, which, for example, has allowed BBVA to introduce new categories of expenses in your application.

One of BBVA's main objectives with this movement is for 'fintech' and 'insurtech' (technology companies oriented towards finance and insurance) to use its code.

At the same time, he hopes that the contributions of the community will allow them to strengthen their algorithms.. “These types of algorithms are not very common in the 'open source' field,” Curro Maturana, Curro Maturana, explained to El Mundo.