An applied-AI studio. We build decision tools from real data and publish the honest numbers.
Our first product reads the snow that fell months ago and estimates the probability of spring flooding for six river basins, calibrated, with confidence intervals, through an API and a mobile app. Trained on 30 years of observations and scored the honest way.
Open Tasqyn →One call returns the next few months of flood probability for a basin, fetching live weather for you. Six basins, a one-to-six month horizon.
A naive split made the model look perfect. We threw that number out and report the walk-forward score instead, the one we'd actually defend.
One rule: build tools from real, local data, and ship the number you'd stand behind, not the one that looks best on a slide.