FitsFlow: A Cyberinfrastructure, Browser-Based, Cloud-Native Platform for Machine Learning–Ready Processing of Solar FITS Data

FitsFlow is a browser-based platform developed by the author to streamline the exploration and annotation of Solar Dynamics Observatory (SDO) FITS images through a fully integrated, cloud-native environment. The system connects heliophysics data services from Joint Science Operations Center (JSOC), Helioviewer, and the Heliophysics Event Knowledgebase (HEK), enabling users to parse FITS headers, align event times, and automatically retrieve associated imagery and metadata. The backend, deployed on Amazon Web Services (AWS) Lambda and Elastic Cloud Compute (EC2), handles on-demand processing and delivers all outputs through a lightweight web interface. FitsFlow produces structured, machine learning–ready outputs, including support for the Advanced Scientific Data Format (ASDF). Each session allows up to 170 MB of fits data and the results can be downloaded in bulk as a ZIP file containing: header metadata in JSON, pixel data in CSV and NumPy formats, colorized PNG images from Helioviewer, grayscale PNGs from JSOC, and HEK metadata in JSON format. These outputs are designed to support reproducible and interpretable ML workflows for classifying, segmenting, and forecasting solar events, laying the foundation for cyberinfrastructure that connects solar surface activity to radiation impacts in low Earth orbit (LEO). FitsFlow represents the first in a planned suite of “KISS” tools (Keep It Simple, Scientist) aimed at lowering the barrier to entry for machine learning in heliophysics. With browser-native visualization, downloadable structured examples, zero-install access, and no login required, FitsFlow broadens accessibility for researchers, educators, and citizen scientists working with solar data.