About Odeai
We develop fast-refresh, observation-driven weather models that combine radar and satellite data with physics-informed AI, for short-range precision.
The company
Odeai develops end-to-end forecasting models. By controlling the entire pipeline and leveraging AI, we run faster execution cycles without relying on intermediary actors. That agility is a clear performance advantage, particularly for short-term forecasts.
- Forecast horizon
- 3 h
- Time step and refresh
- 10 min
- Grid spacing
- 4 km
- Inputs: NOAA MRMS, GOES
- Radar + satellite
How the pipeline works
Conventional prediction passes observations through a chain of processing steps, each adding error and delay. We run the chain ourselves, from raw observation to forecast, in three steps.
- 01
Every sensor, read together
One encoder brings radar, satellite imagery, and other sensor data into a shared representation the model reasons over.
- 02
Forecasts that stay sharp
The model predicts the future in its internal representation instead of pixel by pixel, which keeps forecasts sharp at longer lead times.
- 03
New data in minutes
New observations flow in continuously and refine the model state, so a new forecast is ready minutes after the data lands.
Leadership
Paul Bonnet-Eymard
Chief Executive Officer
Product strategy and business development for industrial markets.
Thibaut Cassard
Chief Technical Officer
Deep learning and atmospheric science, applied to operational weather forecasting.
Thomas Bouvier
Head of AI Infrastructure
High-performance computing and AI systems, applied to running the forecast pipeline in production.
Scientific board
Dr. Louis Thiry
Scientific Advisor
AI research applied to atmospheric physics. PhD (ENS), École Polytechnique.
Ruben Sasportes
Energy Markets & Partnerships Advisor
Energy strategy and market development. CentraleSupélec.
Jacques Boitreaud
Growth Strategy Advisor
Entrepreneur, co-founder of Chai Discovery. École Polytechnique, McGill University.
Make the call with the forecast in hand.
Tell us the decision you make and where. We will show you the forecast on your own past events.