Our blog
Our latest news and announcements.
Our latest news and announcements.
Raystacks organize radar observations into aligned arrays for training across sweep and volume boundaries. We explain folding, batching, and how radar observations become training samples.
Silurian Team
September 16, 2026
An open-source NEXRAD Level 2 reader with a Rust core and a Python xarray compatible interface.
Silurian Team
September 10, 2026
When rime ice brings down a transmission line, it's already too late. Traditional weather models can't see it coming—they forecast regions, not assets. Our collaboration with Hydro-Québec validates that a weather foundation model like GFT can deliver day-ahead rime-ice alerts with 0.72 average precision, reducing expected dispatch costs while providing operators with actionable 72-hour warnings.
Silurian Team
October 24, 2025
Silurian and Hydro-Québec partner up to train Silurian’s Generative Forecasting Transformer (GFT) on utility-grade grid observation data to capture asset-level details and provide early warnings for catastrophic icing events. These results were recently presented at a CIGRE 2025 keynote.
Silurian Team
October 8, 2025
Talk to us about a deployment. See proof across energy and government.
A team of researchers and engineers from:
