Meet Mario López, Data Engineer at Mercadona Tech

In conversation with Mario Lopez about his career in Data Engineering, the ups, and downs of his professional journey. From the Ph.D. student to the crucial team member in the tech revolution of Mercadona.

JS: What is your background, and how did you get to work in Data?

ML: I studied Telecommunication Engineering, and after graduating, I got offered to stay at the University of Cartagena to proceed with my doctoral work. I was researching the application of Decision Theory mechanisms (Markov Decision Processes, Game Theory, Multi-armed Bandits…) to share radioelectric spectrum among wireless devices in optimal ways dynamically, in other words, finding ways for you to download cat videos faster. My focus was machine learning, not precisely to the point you apply it in business, but it went into a specific data direction rather than telecommunication.

During that period, I learned a lot about the importance of soft skills that have accompanied me. After defending my thesis, I started to work in tech startups in the software development teams. Although I was part of the engineering teams, working with software because of the products we were working on, I learned a lot and was soaked with data since the beginning. I discovered what I liked most was data engineering. We built a reasonably advanced tool at the time for data analysis and science work and visualization (kind of a Jupyter Notebook on steroids), and another data engineering team was creating a tool for retail to analyze vast amounts of data. I led them and proudly looked for my place in the tech world.

JS: What made you move to Valencia? I know you moved from Madrid.

ML:It was because of Mercadona Tech. My family is from Murcia, so it was also crucial to be closer. The primary reason was the team I encountered when I was interviewing and the CTO, Fernando Diaz, and CEO, Juana Roig. I noticed how careful they were while choosing Talent, and I liked the idea of working with people from whom I can learn a lot.

Valencia has a different rhythm of life passing by. I love Madrid, but I came to Valencia, and I appreciate the sunshine almost every day, the real estate market and the beach… (laughs). This city gives you some relief sensation, over 20 degrees in December. It rarely happens in Madrid! Here, it’s something usual.

JS: Mario, as the Data Team leader, tell me what impact you have on Mercadona Tech?

ML: Woah! We have a vast impact! For starters, we help provide the foundations for making sure all the informed decisions Mercadona takes are based on correct data. The data is how it should be, to facilitate it to Product Management so they can analyze it. And then, the business decision has a direct effect on the jefe and our project. The other part is our Data Science that the advanced application can build thanks to the data we obtain.
We had payment issues, and our data apps are made to save on costs or possible fraudulent transactions, which are common in e-commerce. We improve the payment management for our clients; we make it safer for the business and the jefe. We are present in many areas of our operations, from detecting fraudulent purchases (protecting our Jefes from stolen credit cards) to recommending products to spice up everyday life, different forecasts to satisfy our Jefes´s demands.

JS: What tools are necessary for the Data team?

ML: Is ‘curiosity’ a tool? No, more than anything, what is working best for us, by far, is writing. I believe that it is required for any non-trivial work, but specifically in data. In Data, attention to detail is crucial. Many nuances arise from the many stakeholders we work with, the different assumptions we need to make, all the required moving parts for a data product to work. Nothing captures this better than the written word. So, we are in love with Notion. There we can reach clarity together. The feedback is essential, and it facilitates work a lot. And for wide sharing across the company, we use Confluence. On the hardcore techie side, our favorite tool is DBT, a data transformation tool that enables data analysts and engineers to transform, test, and document data in the cloud data warehouse.

JS: How do you get Good Quality Data?

ML: For that, we use DBT. We make sure our data sources and our data transformation are both tested. Also, we genuinely believe that Good Quality Data cannot be achieved without each team at Mercadona Tech assuming ownership of their data and a product mindset applied to it. It’s still a challenge for our Data Engineering team, and we work on it currently.

JS: What will be the responsibility of a new Data Scientist joining the team?

ML: There are a lot of things to improve in our e-commerce. We are not a mature team yet. Despite some successes, It has taken us a while to build a foundation we like in culture, values, relationships with other roles and groups. I think we are now in an excellent place to increase our pace. And for that, we would like to benefit from the experience of a senior hire.

JS: How would you describe the collaboration between Data Engineering and Data Science?

ML: Everybody hates the Thinker/Doer model. Ideally, we would like to go two steps ahead of our data scientists and create the proper abstractions to own their solutions end-to-end. We have intense collaboration focused on the task until we get there, with no handoffs and blurry lines around our roles.

JS: I wanted to share with you this passage that I found interesting, and would like to hear your thought on it:

Having a first hire be a data scientist is like hiring a chef but not having an efficient way to get ingredients. They will spend more of their time gathering ingredients. If you hire a distributor to bring you the food that solves the travel problem but chef still needs to prep the food. An analytics engineer (or SQL developer or BI developer if you like) can assist in prep of the food like a sous chef. At that point the chef (scientist) could hopefully spend most of their time creating meals instead of prepping. In the end it is a supply chain problem. People are realizing they have a problem with ingredients before they have a problem with cooking. — by flerkentariner

ML: It’s a typical mistake when it comes to data teams! For a Data Scientist to work efficiently and for product managers to make data-informed decisions with a guarantee, there is a structure that needs to be built by data engineers.

JS: Mario, how do you explain your profession to your grandma? How do you tell people out of the tech world what you do at work?

ML: Hmm..I would say the same as a grandpa who was going to the coal mine to dig the coal, and I dig in the Mercadona Online mine to bring clean data. In order to make the information which already exists, make it useful for my colleagues. Earlier, data engineering and data science were also called data mining. There is a lot of materia prima, and my job is to show its value. And as we advance, make the business better, and make one grandma’s life better, so she can easily buy online.

JS: What advice would you give to someone who would like to work in Data?

ML: Don’t get overwhelmed because it can seem that there is too much to learn. That one should focus on knowing the basics and finally that communication is the key.

JS: What’s your favorite Databook?

ML: Everybody Lies: Big Data, New Data, and What the Internet Reveals About Who We Really Are by Seth Stephens-Davidowitz.

JS: Thanks for your time, Mario!

Currently, the Data team is looking for a new team member. Read more if you feel like joining us!

Although we could have called it simply “data engineer”, we didn’t want to stick to the data engineer role title because:

We have chosen a data stack that has severely simplified (for now) our infrastructure: the Modern Data Stack. In our case, extract-and-load pipelines are custom made (probably not for long), data transformation is implemented with dbt, and our workhorse is BigQuery. Thus, in this sense we are probably closer to #software engineers working with #data.

We are in the early stages of data maturity, thus, we don’t need much specialisation. Our current responsibilities are broader than a typical data engineer would have:

- We don’t believe in the thinker/doer model: we want to build tools for data scientists to own their solutions end-to-end and/or direct engineering support/collaboration as needed, but not in the form of handoffs. Thus, in this sense, we are also ML engineers.

- There are no data analysts in Mercadona Tech: we believe highly technical Product Managers better supply those needs. However, they need good tools to do that job: proper data modeling, data quality… In this sense, we are also analytics engineers.

So… what will you do, specifically? You will, among many other things:

* Implement tools and teach other teams on how they should model their data and keep their data quality up.

* Collaborate with our data scientists (Jorge Cabrejas Peñuelas & Rubén Ibáñez Pinillo & Javier Punzano) to develop and maintain the software side of MTech data science solutions.

* Help teams avoid their BigQuery costs to skyrocket.

*Everyone’s favorite: pipelines!

So, any #dataengineer in the room? Apply here!