- Are you well-versed in AI/ML and/or forecasting?
- WMO/WWRP, WCRP, s2sprediction.net organize a prize competition with 15K CHF for the winning team
- Goal: Improve bi-weekly subseasonal-to-seasonal predictions with AI/ML (or try to prove that AI is not needed with other methods)
More info: https://s2s-ai-challenge.github.io/
Users will likely use xarray and dask. What other software I think could be relevant:
- Machine Learning in Python: https://scikit-learn.org/
- Scalable Machine Learning with Dask: https://ml.dask.org/
- Metadata-aware sklearn: GitHub - phausamann/sklearn-xarray: Metadata-aware machine learning.
- Batch generation from xarray datasets: GitHub - pangeo-data/xbatcher: Batch generation from xarray datasets
- Parallel Machine Learning: https://ensemble-learning-models.readthedocs.io
- Metrics for verifying forecasts: GitHub - xarray-contrib/xskillscore: Metrics for verifying forecasts
- Verification of weather and climate forecasts: GitHub - pangeo-data/climpred: Verification of weather and climate forecasts.
- https://keras.io/ https://www.tensorflow.org/ Welcome to PyTorch Tutorials — PyTorch Tutorials 1.8.1+cu102 documentation
My role is to organize the competition from a technical point of view. We collaborate with ECMWF to provide S2S output on their S3 cloud and renkulab.io for evaluation. Moreover, we provide example notebooks, which can be run on renkulab.io, which is git repositories conntected to jupyterhub. Examples still work in progress…
I got this position via the pangeo forum, thanks @chiaral for sharing WWRP/WCRP S2S AI competition - looking for IT expert in python, git and Jupyter notebooks