Status Report: AI/ML

June 2022- October 2022

Unidata Program Center Staff

Areas for Committee Feedback

We are requesting your feedback on the following topics:

To you and your own community, what tutorials or resources for machine learning are the highest priority?

Thank you to everyone who responded to the pre-workshop survey!

Activities Since the Last Status Report

Thomas Martin was hired in July to resume the role of AI/ML Software Engineer
Added support-ml@unidata.edu as a direct support channel
Python resources have been shared on Project Pythia and other openly available github repositories
Started testing GPU (graphics processing units) resources on Jetstream2 for wider adoption.

Machine learning is actively being used for earth systems science and research. Most online training and examples do not use earth system science examples due to specific dataset I/O and pre-processing requirements. We plan on developing training resources to help bridge this gap, while using applicable earth system science examples. Besides tutorial development for scientists who are new to machine learning, python workflow development of tighter integration of xarray datasets to machine learning packages is a high priority. We are also actively exploring where current and future unidata tools (e.g. netCDF, THREDDS, etc.), can be leveraged for machine learning research, training, and deployment.

New Activities

We are looking for more ways to support our community at large. This might include in-person or remote workshops, 1 on 1 mentoring and tutoring, and online aschynorious resources. Expect these activities to increase in 2023. Thomas will also be at AGU (December 2022) and AMS (January 2023) available to discuss potential workshops or collaborations.

Prepared October 2022