Three keys to working effectively on Artificial Intelligence projects
Marta Sanz, COO at BBVA AI Factory, analyses the challenges faced by the BBVA AI Factory team in creating a new, more agile and efficient working model.
On many occasions, the greatest impediments to creating Artificial Intelligence solutions do not lie in the capacity of highly qualified teams, but in establishing an effective way of working between the different professional profiles involved in the life cycle of analytical models. This is one of the main tasks we are currently tackling at BBVA AI Factory. It is a task guided by three concepts: simplify, accelerate and reuse.
My first direct contact with the AI Factory was in April 2020, in the middle of lockdown. I found myself with a team of data scientists who were extremely competent in creating AI models, but who needed to continue to push for common working guidelines in order to deal with the complexity - both organisational and technical - that exists in the Engineering domain. At the end of the day, models and data engines have to be integrated into the various channels of the bank to be made available to our clients. This makes it essential to work together as one team. The best models are those that reach the end user.
One of the main chapters we are starting to work on is precisely to simplify the collaboration model between the worlds of data science and engineering, thus reducing organisational complexity and allowing teams to gain autonomy. It is important to understand the complete development cycle (data preparation, engine development and integration) and the role of each team during each phase, trying to eliminate blockages which are often caused by incorrect management of the dependencies between the different work groups.
Anticipating dependencies allows us to accelerate the delivery of models and data engines and also to parallelise the work of all the teams. Without a doubt, being able to gather feedback on how our models behave in the quickest possible time is a fundamental aspect for us. Our aspiration is to reduce the total time we spend on building analytical models, from the definition phase of the solution for a specific business need, to its implementation in production. In this sense, it is key to measure these times in detail and use this metric as a lever for the continuous improvement of the entire development cycle. In short, it is a matter of reducing downtime, being more efficient and responding better to the expectations of the teams.
The third major challenge in order to make Artificial Intelligence work more effectively is to be able to reuse developments as much as possible. We start from a way of working in which data scientists and Machine Learning engineers work on specific projects in which they have a partial vision of the problem to be solved. We have to integrate all these visions and build global data products that can be reused in the different areas and countries in which BBVA operates. This way, when we design a new data-based product, it can be exported and adapted much more easily. At present, synergies are detected, but sometimes teams face similar challenges without taking into account the previous work and experience of other areas.
With these ideas in mind, as COO (Chief Operating Officer) I join this great AI Factory team, recently set up as part of the BBVA Group, acting as a bridge between the worlds of Data and Engineering. Furthermore, I am fortunate to be doing this alongside Francisco Maturana (Company CEO) and Ricardo Oliver (head of Data Engineering at BBVA), from whom I receive key support and with whom I am fully in tune in the implementation of these working guidelines. In recent years I have worked within the Architecture & Global Deployment team after spending most of my career at BBVA in the field of technical and banking architectures, both in Spain and in different countries in the Americas (Mexico, Peru, Colombia, Venezuela and the United States). Right now I am looking forward to meeting my colleagues from the AI Factory in person, ( health situation permitting) and to thank them for the fantastic welcome I have received.
Before ending this short article, I think it is most appropriate to mention the reason why we want to continue advancing the lines of work mentioned above. Ultimately, our purpose is none other than to squeeze all the potential offered by Artificial Intelligence, whether it be in decision-making, process optimisation or in the creation of products that offer added value to our clients. That is why we are working on aspects such as interpretability and fairness in analytical models. We are convinced that AI will bring significant benefits to society in general and to our clients in particular, supporting them in making the best financial decisions.