Three weeks before haCARthon started, I was supposed to have my head buried in excerpts of Brazil’s Forest Code, validating queries against geo databases of Brazilian municipalities. Instead, I was on a video call trying to convince my four-person team to split into two.

It wasn’t an impulsive decision. It was the conclusion of an entire month immersed in a challenge I didn’t expect to enjoy this much.

R$15,000 was the hook I was missing

I was never much of a hackathon enthusiast. But this time the hook was direct: my partner, a UX/UI designer, got an email from PanicLobster — the event organizer — announcing that in a month haCARthon would take place, with a R$15,000 prize for each of the five winning teams. The incentive sparked my interest enough to go dig into the proposed challenges.

A mutual friend, a public administrator who’d already taken part in other hackathons with her, heard about the event and brought one more person: a frontend developer. That’s how the group formed — me, my partner, the friend, and the frontend dev — not yet knowing that, three weeks later, we’d become two pairs.

Nobody in the group understood CAR

With only a month until the event, the first problem wasn’t technical — it was alignment. Cadastro Ambiental Rural (Rural Environmental Registry), the Forest Code, official geospatial reference databases: I could barely decode the acronyms, and the rest of the group was in the same boat. I decided to compile everything I could gather into a single place, built a Docusaurus app with that material, and put it online through my homelab , so everyone could discuss the challenges starting from the same context.

It worked. After several late-night meetings to sketch flows and personas, by the first week we already had roles defined: the UX/UI designer and the frontend dev would turn the flows into screens and prototypes, the public administrator would handle the pitch narrative, and it fell to me to ground the technical feasibility of the solution.

That’s how Carla was born — a name picked as a joke, but that later made too much sense to change. The idea was simple to explain and hard to implement in practice: a guided, in-house, open-source chat interface for citizens to declare and rectify their CAR without depending on a paid intermediary. Instead of asking the rural landowner to draw the property polygon from scratch, Carla suggests a boundary based on what they’ve already provided, automatically highlights Legal Reserve and Permanent Preservation Areas, and never asks again for data it already received — it closes each step with a summary, not field by field. On the other side, a portal for the environmental analyst, with a prioritized queue and a supporting dossier.

The documentation kept getting updated as the ideas matured, and we all knew the project would still change during the three days of the event — there we’d have access to mentors and more concrete decisions.

Why double down

Since technical feasibility fell to me, I stepped into a completely new world: Forest Code law, georeferenced databases, QGIS, geo-mapping techniques, revisiting linear algebra applied to vector calculus, using drones and satellite imagery to map territory.

I was enjoying what I was learning way too much, and we still had three weeks until the event. I brought an idea to the group that wasn’t in the plan: also tackle Challenge 2 — the most technical of the three, precisely the one I figured fewest teams would be willing to take on.

Except there was a rule that complicated everything: each team could only submit for one challenge. If we wanted to deliver two projects, we’d need to become two independent teams — and split a group of four people into two pairs.

We decided to take the risk. I put together a second reference document, along the same lines as the first, to ground this second front. I even considered calling the solution “Carlos,” but it would sound too much like Carla — we named it GeoHub Brasil. The proposal: ingest and normalize eight federal geospatial databases (SNIF, ANA, ICMBio, FUNAI, IBAMA, INCRA, IBGE, INPE) into a single place, and run automatic validations — overlap with indigenous land, conservation unit, IBAMA embargo, Permanent Preservation Area and Legal Reserve calculation — that today an analyst has to manually look up across eight different systems. The deliverable is a preliminary geospatial assessment in under 60 seconds, not a ready-made registration: the user is the analyst and the data manager, not the rural landowner.

Two deliveries, three days, four people

haCARthon started at 2pm on a Friday, and we arrived anxious. Three intense days, emotions running high — a good chunk of the time discussing technical decisions as four people who, in practice, had been laypeople on the subject for barely a month. Mentorship was what gave us the most direction in those moments: alone, we spun our wheels; with an expert mentor, discussions turned into decisions.

We split the pairs the way the previous weeks’ work had already been suggesting: the public administrator and I continued with GeoHub Brasil, the UX/UI designer and the frontend dev continued with Carla. Each pair, alone, had to deliver what would previously have been the work of four — including the two videos per project: prototype and pitch. Splitting the team multiplied our chances of a good ranking, but it also multiplied by two the work each pair had to do in the same three days.

The Chinese tool that saved the last night

By the end of the second day, still needing to record GeoHub Brasil’s pitch video, I stumbled on GitHub onto a tool called MoneyPrinterTurbo , created by harry0703 : a Chinese library that assembles videos from a script, combining short clips from stock video banks. I cloned the project, ran it locally to evaluate it, and saw it could be used outside its original use case — exactly what I needed to turn the pitch script into a video without relying on manual editing.

Using the tool in an unconventional way, I ran into a few bugs. I opened an issue, which ended up accepted as a fix in the project . I spent the last night adapting a Chinese open-source project to my specific context, and in the end managed to generate GeoHub Brasil’s pitch video with the tool.

What was left after the ranking

We submitted both projects with just minutes to spare before haCARthon ended — documentation updated, everything published, waiting for evaluation. Days later the results came out: among more than 230 registered groups, Carla placed 130th, GeoHub Brasil 65th. Nowhere near the podium.

Even so, the risk was worth it. What’s left is a portfolio of two solutions I now show off with pride, a handful of new technology on my résumé, plenty of college material recalled under pressure, and new people I met along the way. The ranking was the criterion the organization used to evaluate us — it wasn’t the only one I used to evaluate whether it was worth it.

Carla documentation: https://guirondon.dev/portfolios/carla-hacarthon
GeoHub Brasil documentation: https://guirondon.dev/portfolios/geohub-hacarthon

Carla prototype: https://www.youtube.com/watch?v=pFvPlxF-FXs
Carla pitch: https://www.youtube.com/watch?v=JHjPdq5rZw8
GeoHub Brasil prototype: https://www.youtube.com/watch?v=PQTDf4yG5Bw
GeoHub Brasil pitch: https://www.youtube.com/watch?v=OK5n_thwBoU

If there’s one thing that stuck after closing the laptop that night: splitting the team in two wasn’t about doubling our chances of winning. It was about accepting that we could learn twice as much in the same timeframe, even knowing each pair would have half the hands to get the work done.