platoseedplatoseed
Foundation models to simulate Earth, starting with weather.
Backed by Y Combinator · AI Grant
Silurian is building foundation models for simulating Earth, starting with weather. From assessing the risk of wildfires to predicting the energy grid load, we provide an infrastructure layer for our planet. Our frontier models push the boundaries of what can be simulated on Earth and improve decision making across vital sectors including energy, insurance, agriculture, and logistics.
Silurian AI builds foundation models to decode and forecast Earth's physical systems, starting with weather, and offers tailored, AI-powered regional forecasts and deployment options for organizations handling Earth data.
Build a tailored foundation model that maximizes the value of asset and infrastructure data feeds to generate high-resolution, rapidly refreshing regional weather forecasts and Earth-system insights; offers deployment talks and demos, and provides bespoke solutions for energy, government, and other sectors.
Who it’s for: Organizations with asset or infrastructure data needing high-resolution weather or Earth-system insights (e.g., energy, government, utilities, large enterprises).
Has published research (Aurora foundation model) and offers demos/deployments; active on pricing page and mentions deployments, indicating product-ready and customer-facing activity.
Jayesh is Co-founder and CEO of Silurian AI. Previously he was the Head of AI at Poly Corporation. He led the development of the first foundation model for weather & climate as researcher at Microsoft and was the co-author of Aurora. Jayesh holds a Ph.D. in CS from Stanford.
Chief Scientist of Silurian. I was previously a Senior Researcher at Microsoft Research where I co-lead the development of the Aurora foundation model. During my PhD at Cambridge, I've also spent time in top AI industry labs like Google Brain, Google X and Twitter Cortex.
AmazonNikhil is a co-founder and chief engineer at Silurian AI. Previously he was a software engineer at AWS SageMaker and prior to that he dropped out of his PhD in applied mathematics where he was working on problems in fluid dynamics.
Foundation models for simulating Earth, starting with weather.
Silurian launches GFT, a 1.5B Generative Forecasting Transformer that simulates global weather up to 14 days ahead at ~11km resolution, claiming up to 30% accuracy improvements over NOAA/ECMWF. The API provides location-based queries and targets verticals like energy, climate, and agriculture.
Building a nervous system for the atmosphere
Foundation models for time series
