Sarus

Sarus is the privacy layer for analytics and AI that sits between data sources and analysts, enabling work on sensitive data without exposing it. It enforces privacy guarantees on every query using differential privacy, synthetic data, and privacy-preserving query rewriting, while supporting on-premises or public cloud deployments. This approach helps research, analytics, and AI use cases unlock data value securely and compliantly.

• Connectors: natively connects to databases, data warehouses/lakehouses, and cloud storages without moving data.
• Privacy Policies: define outputs (synthetic data, aggregates, differentially private outputs) and audit all processing.
• Audit Logging: full query auditing for compliance.
• Synthetic Data: high-fidelity, DP-trained synthetic data that preserves distributions.
• Analytics & BI: SQL API and BI connector support for common tools (PowerBI, Tableau, Metabase).
• Machine Learning: Python SDK to run remote data processing with privacy enforcement (DP) when needed.
• Action Upon Data: push query outputs to external endpoints without exposing personal information.
• Privacy-Preserving DP, Privacy-first Query Rewriting, and Output-Level Control to ensure privacy across complex pipelines.

• Strong privacy guarantees: differential privacy, synthetic data, and on-the-fly policy-enforced query rewriting.
• Data never leaves data owners’ control: analysts work on remote data with no direct access to raw data.
• Compliance-friendly: detailed audit logging and streamlined privacy/compliance workflows.
• Data value across silos: unlock data insight across systems without compromising privacy.
• Supports AI/ML workflows and LLM usage with privacy guarantees.
• Flexible deployment: on-premises or in your public cloud (Docker or Kubernetes) with broad source connectors.