SpiralDB

Spiral is a storage and query engine turning raw robot data into curated, versioned datasets for machine learning. It lets robotics/physical AI teams query multimodal data (video, audio, images, sensor traces, tabular) directly from cloud storage without ingestion, then investigate, enrich, and curate exact dataset slices with version control and time-travel. GPU-optimized loading streams curated results into training pipelines for teams managing massive, fast-iterating data volumes.