Security & compliance

Designed so the worst case is small.

Tokenization, key isolation, and least-privilege access are not bolted on — they are the substrate. A compromised analyst account leaks tokens, not customers.

Tokenize-at-ingest

Sensitive fields never land on the lake in cleartext. If an analyst dumps a table, they take tokens with them.

HSM-backed vault

Token keys live in FIPS 140-3 Level 3 HSMs. They never leave the boundary, and you can bring your own.

Zero-trust access

mTLS service identities and short-lived capabilities mean a stolen credential expires faster than an attacker can pivot.

Tenant isolation

Compute pools are not shared. Storage namespaces are not shared. Key material is not shared. Period.

Tamper-evident audit

Every read, reveal, policy change, and key rotation lands in a hash-chained ledger that can be replicated off-platform.

Compliance, day one

SOC 2 Type II, ISO 27001, HIPAA, PCI DSS, and GDPR controls mapped to specific platform behaviors — not just policies.

Threat model

We assume the analyst is the breach.

The traditional lake assumes everyone with a query client is well-behaved. rjbase.io assumes the opposite: every query is a potential exfiltration attempt, and the system is designed so that the worst outcome is a pile of meaningless tokens.

  • Compromised credentials cannot detokenize without a fresh justification
  • Bulk reveals trigger rate-limited approvals and notifications
  • Engine-level masking — no client-side enforcement to forget
  • All inter-service traffic is mTLS; nothing inside is “trusted”

Certifications & attestations

SOC 2 · Type II
ISO 27001
HIPAA
PCI DSS
GDPR
CCPA

Reports available under NDA. Email security@rjbase.io.

Now in private preview

Tokenize your data. Keep your governance.

Talk to our team about an evaluation cluster. We typically have engineering partnerships running inside two weeks.