AI for climate hackathon

In 2021, the University of Osnabrück hosted a Hackathon on using deep learning and machine learning for climate and cryosphere forecasting. The Hackathon consisted of three days of hacking, and a final presentation.

Seq-to-seq architecture
Web app example

Our Team, Snow Clue, placed first with an innovative approach of using convolutional sequence-to-sequence LSTMs for short to intermediate snow cover forecasting and took the first place in that category. I developed the entire deep learning pipeline, including data preparation, validation and optimization. My team developed a web application to visualize results and make them easily accessible.

Prediction results