01 / Evidence
Measure before claiming.
Results are tied to a named checkpoint, a fixed period, a defined grid, and an explicit metric.
About Editable.ai
Small team. Measurable work.
Editable.ai is a Pakistan-based weather research and engineering studio. We build regional forecasting systems, severe-weather hindcasts, and traceable weather-evidence pipelines.
Research lead
Syed Sabbih Haider Shah
Weather and climate machine learning
Geospatial systems
Scientific computing
Sabbih leads the development of forecast and severe-weather models spanning atmospheric data preparation, distributed training, multi-decade hindcasts, spatial evaluation, and event-set construction. Current work uses ERA5, URMA, MRMS, NOAA storm reports, and operational forecast guidance.
Email SabbihWorking principles
How the research is run.
Good scientific engineering is mostly discipline: define the comparison, keep provenance intact, and make limitations easy to find.
02 / Provenance
Keep the source visible.
Model configurations, hindcast files, label logic, and figure inputs remain traceable through the pipeline.
03 / Limits
Say what does not work.
Calibration gaps, resolution changes, reporting bias, and weak metrics are part of the result—not footnotes to hide.
04 / Release
Build toward reuse.
OpenSCS is organized around reproducible labels, yearly hindcasts, evaluation scripts, and eventual benchmark releases.
Contact