Aeropuertos Argentina

Where they were

Aeropuertos Argentina is the company that manages 35 airports in Argentina, so if there’s one thing they have (and plenty of), it’s data. Because of their unique characteristic, there’s a number of systems that exceeds what’s typical in a company: parking, chatbots, marketplace, sensors, surveys and CRMs. Each one with its own source, logic, format, and characteristics. When they wanted to build the passenger journey, they discovered they didn’t have the necessary visibility into their data or into the possible contact points between these systems.

What we found

The problem was visibility and governance. There wasn’t a complete map of data sources or how they related to each other, plus the ETL processes had years of technical debt, with undocumented logic in M and Power Query, duplicated tables, and data in the wrong layers of the architecture. So, before building something new, we had to understand and organize what already existed.

What we built

First, we did a complete survey of data sources, systems, and connections across the 35 airports. Once the documentation of domains, glossaries, and catalog was validated, we moved forward with the concrete migration: from Power Query to PySpark with star schema modeling, documentation of logic that wasn’t written down anywhere, and cleanup of orchestrators in Azure Data Factory.

What they can do now

The main pipeline reduced its execution time to a third of what it used to take, with a significant reduction of errors in production. Now the team has a clear map of their sources, a governed catalog, and documented processes they can maintain and extend without depending on who originally wrote them. The foundation is set for the AI initiatives the team is exploring.

Why they came back

Two separate contracts: the first was diagnosis and strategy, and when they saw the result, they trusted us for the execution.