Security posture
sitkastack handles client data with the same discipline I bring to the AI systems I build for clients. This page describes my current practices, attestations, and what's available on request.
Last updated: May 2026
Compliance status
- SOC 2 readiness assessment: complete. Available on request.
- SOC 2 Type 1 attestation: in progress. Audit booked for Q4 2026, report expected Q1 2027.
- SOC 2 Type 2 attestation: planned, Q2 2027.
- Cyber liability insurance: $1M coverage. Certificate available on request.
Today's practices
Identity and access
- Single sign-on across all client-data-handling systems (Google Workspace)
- Multi-factor authentication enforced on every system
- Hardware security keys (FIDO2) for sensitive access
- Quarterly access reviews documented
Device security
- Mobile device management on all work devices
- Full-disk encryption (FileVault) enforced
- Screen lock under 5 minutes
- Automatic OS updates within 30 days of release
Data handling
- Per-engagement data handling agreements before any client data is exchanged
- PII detection and redaction by default in any AI pipeline that processes client data
- Synthetic test data used in public Framework reference implementations; no real client data ever appears in public artifacts
- Audit logging on all production AI systems I build for clients
AI-specific practices
- Prompt injection mitigation built into every production agent I ship
- Confidence-gated routing with human-in-the-loop for any decision affecting client operations
- Model cards documented for every model deployed in a client environment
- Token usage and cost tracking on every production deployment
Vendors and subprocessors
- Anthropic (Claude API)
- OpenAI (where used per client requirements)
- Google (Workspace, Gemini API)
- Postgres hosting providers (per-engagement, varies)
- Cloudflare (DNS, edge)
Full subprocessors list available on request, including data handling provisions and DPAs.
Incident response
- Documented incident response procedure
- Quarterly tabletop exercises
- Client notification commitments per engagement contract
Frameworks I align to
sitkastack engagements ship governance artifacts mapped to twelve regulatory frameworks and standards across detailed mappings and engagement-supported categories.
View the full frameworks index
Privacy and data protection frameworks applied as the engagement context requires:
- PIPEDA (Canada)
- GDPR (where applicable)
On request, available under NDA
- Subprocessors list with DPA references
- Data flow diagrams for typical engagement patterns
- Incident response procedure
- Quarterly access review log
- Cyber liability insurance certificate
- SOC 2 readiness assessment document
- Per-engagement security questionnaire response
Contact: robyn@sitkastack.com
For procurement teams: most security questionnaires I receive can be answered by combining this page with the documents listed above. Send your questionnaire and I'll typically respond within 3 business days.
