Nowcasting-in-a-Box (NiB) is a portable, modular and containerized AI nowcasting framework designed to address the capacity gaps in weather nowcasting (0–6 hour predictions) faced by WMO Members worldwide, particularly NMHSs with limited scientific development capacity, computing infrastructure, observational coverage or specialized AI/meteorological expertise.

• Lightweight Architecture: Packaged as a Docker/containerized or GitHub source checkout solution that runs on low-cost hardware. NiB will support different deployment profiles, including CPU-based, consumer-GPU, institutional-GPU and cloud environments.
• Minimal Local Requirements: Optimized for low-bandwidth internet environments; supports offline operation for data ingestion and model execution, with sync capabilities for cloud-based data sharing (if available).
• Local Data Loader: Provides data decoding, format conversion, quality control and preprocessing for local observational datasets.
• User-Friendly Visualization: Intuitive dashboards for real-time nowcasting product display (e.g., rainfall intensity maps, lightning probability, radar echo extrapolation) —compatible with local NMHS warning systems and designed for forecaster/decision-maker usability.