Plan overview · May 2026

Arawave is a Paraguay weather AI project. Here's where it stands and what's next.

I built a probabilistic precipitation forecasting system for Paraguay's agricultural sector. The model is done and it works — it beats the industry-standard baseline by 25.7%. The remaining work is not technical; it's getting the right Paraguayan institutions to look at it. This is the plan for that.

Quick read

If you only have two minutes:

The project — what was built.
Goals — what I'm actually trying to do, in stages.
How to help — specific things that fit the current stage.

If you want the full picture, read all eight pages — each is a 1–2 min read.

The headline

vs. industry baseline

+25.7%

RMSE improvement on 10-day precipitation forecasts

Coverage

25 km

grid resolution over the soybean belt

Validation

60 dates

against ERA5 + CHIRPS reanalysis

Out-of-pocket so far

$60

over 2 months · plus $60 in Modal credits = $120 total compute

Live demo

The technical demo is online at stellar-pika-fa02e6.netlify.app. It shows scorecards, per-event maps, calibrated uncertainty, and the honest caveats. Mobile-friendly.

What this site is

This is a plain-language overview of where the project stands and what I'm doing next, written for someone who isn't deeply involved in it. The eight pages cover what was built, where the gaps are, and the operational plan for the next three months.

If you have advice, questions, or a useful introduction — I'd love to hear it. The How to help page lists specific things that would actually move the project forward.

Project lead: Kevin Hill.
Meteorological collaborator: Fran (ex-DMH Paraguay).
Status as of: May 2026.