From software to data: how we create the tools we need to succeed

Every commercial flight in Argentina, tracked in high-frequency, on a $0 budget.

This wasn’t a client request. It was a personal mission. When I joined Blueprintdata as a Data Engineer, I realized that understanding this world meant more than just reading documentation: it meant building something that shouldn't be possible with the technical overhead and manual barriers I saw everywhere.

Transitioning from Software Engineering to Data Engineering felt a bit like landing on another planet. Vining from the software world, I thought I knew what to expect: Python, APIs, and the assumption that SQL was just for "getting data."

I was wrong. The complexity of the SQL models we handle at Blueprintdata is fascinating. It’s not just about queries; it’s about how you model, measure, and define information in ways I hadn’t imagined. But reaching that "interesting part" felt impossible at first. The Modern Data Stack was fragmented, and I spent my first days fighting the "Dark Side" of data: infrastructure boilerplate, manual service wiring, and the repetitive manual tasks of starting from a blank page.

The Mentor's Tool: Blueprintdata CLI

During my onboarding, I expressed this frustration to Manuel Ponsa and Marco Porracin. They didn't just give me a manual; they pointed me toward a solution we were building to eliminate these bottlenecks: the Blueprintdata CLI.

The CLI was designed to be the "lightsaber" for a data hero. It standardizes our architectures, automating the 90% of plumbing that usually eats up the first week of any project. Instead of wiring together Meltano, dbt, and Cloud storage manually, you could now bootstrap a production-ready stack in seconds.

Helping build this tool while I was still learning the ecosystem was a game-changer. It taught me that in the data world, the best engineers don't just move data; they build the machines that move it.

Testing the Mission: The Flight Tracker

Once we had a version of the CLI we believed in, Manu gave me a challenge: "Build something with real value that we can visualize." Combining my software background with my new data skills, I decided to track every commercial flight in Argentina.

Using the CLI, I started the project with a single command:

npx blueprintdata-cli template

In 30 seconds, I had my foundation:

Orchestration: Meltano for automated data syncs.
Transformation: A dbt project structured for the complex modeling I mentioned earlier.
Automation: GitHub Actions to deploy changes safely.

Solving the Ingestion Puzzle: tap-airlines-arg

The data source was the real obstacle. Aeropuertos Argentina 2000 manages the traffic, but there was no API or "official" pipe.

I used the Meltano Singer SDK to build tap-airlines-arg, creating a robust contract between a "closed" source and our warehouse. I published it on PyPI so the next engineer could skip the struggle. This was the moment ingestion stopped being about "scripts" and started being about software principles.

The Result: A $0 High-Frequency Dashboard

By combining the CLI’s foundation with the custom tap, I built an architecture with zero fixed costs:

Ingestion: tap-airlines-arg pulls data in high-frequency syncs.
Warehousing: Raw data lands in BigQuery’s free tier.
Modeling: dbt deduplicates and structures the logic of flights.
The Static Trick: Final models export as JSON files to Google Cloud Storage.
UI: A Next.js frontend renders the movements at infoaeropuertos.ar.

Real-time flight dashboard for all commercial flights in Argentina

The Realization: Data is Just Software

What surprised me most is that data is just software.

The same principles (version control, testing, automation) are what make a data stack "modern." Being a builder of the tools (the CLI) while being a user of them (the flight project) is the only way to master this field.

If you’re a software engineer feeling overwhelmed, my advice is: find a dataset you care about and build the tool you wish you had. That’s how I ended up tracking every plane in Argentina.

What's Next

We’re continuing to iterate on the Blueprintdata CLI to make the transition into data as smooth as possible for everyone else.