Beyond the Laptop
neuros-cloud is what lets neurOS run as real infrastructure — streaming data across a message
bus, landing it in a lakehouse, and training across machines or institutions.
package map
neuros-cloud — streaming, storage, and federated training
The distributed backbone: message-bus streaming, lakehouse storage and export, federated learning, and LSL sync for real deployments beyond a laptop.
Streaming
Kafka / Redis / ZeroMQ
message-bus ingest
LSL sync
lab-streaming-layer timing
Storage & export
Iceberg ETL
lakehouse streaming job
Petastorm / WebDataset
training-data export
Database
metadata and provenance
Distributed
Federated client + aggregator
privacy-preserving training
Label Studio
annotation integration
What It Comprises
- Streaming / ingest — Kafka, Redis Streams, and ZeroMQ publishers, a no-op writer for dry runs, and LSL (lab-streaming-layer) sync.
- Storage & export — an Iceberg streaming ETL job, Petastorm and WebDataset exporters, and a metadata/provenance database.
- Federated — a federated-learning
clientandaggregatorfor privacy-preserving training. - Annotation — Label Studio integration and a cloud pipeline runner.
How It Differs
The other packages assume data is already local; neuros-cloud handles movement, persistence,
and distribution — the parts that make the difference between a demo and a deployment. It is the
plumbing beneath the Constellation multi-modal demo.
Part of the neurOS-v1 platform.