Security & Data Protection
How we handle client data, AI governance, confidentiality, subprocessors, data retention, and DPA requests. Last reviewed: April 2026.
1. Our Commitment
SARVAYA (operating from Jaipur, Rajasthan, India) builds websites, AI agents, and automations for clients across India, the United States, the United Kingdom, the EU, and the UAE. Every engagement is governed by a mutual NDA, a written Statement of Work, and - where applicable - a Data Processing Agreement (DPA).
This page is a living summary of our posture. It is not a substitute for a signed contract; it is the reference that saves you 30 minutes in a security questionnaire.
2. Confidentiality & NDAs
- Default NDA: A mutual NDA is signed before any commercially sensitive material is shared.
- Client-provided NDAs: We counter-sign your NDA as standard when provided.
- White-label engagements: Attribution is blocked end-to-end. No client-facing material, repo metadata, or deployment footprint references SARVAYA without written permission.
- Staff: All contributors are under individual confidentiality agreements covering source, data, and client identity.
3. Data Processing Agreement (DPA)
A SARVAYA DPA template is available on request. It covers:
- Roles (Controller / Processor) and lawful basis
- Categories of personal data and processing purposes
- Sub-processor use, notification, and right to object
- Transfer mechanisms (SCCs for EEA / UK transfers where relevant)
- Security measures (Section 5) and breach-notification timelines
- Data subject rights assistance and deletion on termination
We also counter-sign client-provided DPAs. To request a copy, email dev.sharma@sarvaya.in with subject line "DPA request".
4. AI Governance
We use modern AI tooling to deliver faster and cleaner work. We apply the following rules so that the speed never comes at the cost of your data:
- Zero-retention APIs by default: Claude API (Anthropic) and OpenAI API are used with zero-data-retention modes where available. Outputs are not used for model training.
- No production secrets in consumer chat UIs: API keys, PII, and proprietary datasets are never pasted into ChatGPT, Claude.ai, Gemini, or similar consumer chat surfaces.
- Human review: AI-generated code is read, tested, and merged by a human engineer. Nothing ships AI-to-prod unreviewed.
- Model inventory: For regulated engagements, we provide a written list of models used, where they run, and what they see.
- Prompt isolation: Client-specific prompts and retrieval corpora are scoped per-project; no cross-client embedding stores.
5. Security Measures
Access
- Multi-factor authentication on GitHub, Google Workspace, Vercel, Cloudflare, and all model-provider accounts.
- Principle of least privilege. Client-production credentials are scoped and rotated on project close.
- Hardware-backed SSH keys and managed secret vaults for production deploys.
Transport & Storage
- All hosted surfaces served over TLS 1.2+ with HSTS.
- Repositories are private by default. Production environment variables live in managed secret stores, never in repos.
- Laptops are full-disk encrypted with automatic locking.
Operational
- Dependency pinning and automated vulnerability scanning on all production repos.
- Pre-deployment checklist includes a security and PII review.
- Incident response: a confirmed data incident is communicated to the client within 72 hours, aligned with GDPR Art. 33 expectations.
6. Subprocessors
Our standard subprocessors - the vendors we rely on to operate - are listed below. A project-specific list is shared before contract signing. Clients are notified in writing 30 days before any material change.
| Vendor | Purpose | Data accessed | Region |
|---|---|---|---|
| Vercel | Web hosting & edge | Site content, logs | Global (US / EU edge) |
| Cloudflare | DNS, CDN, WAF | Request metadata | Global |
| Google Workspace | Email, Docs, Drive | Business comms | US / EU |
| GitHub | Source control | Source, issues | US |
| Anthropic (Claude) | AI reasoning & code | Scoped prompts (no retention) | US |
| OpenAI | AI reasoning & code | Scoped prompts (no retention) | US |
| Google (Gemini, GA4) | AI + site analytics | Scoped prompts; anonymized analytics | US / EU |
| Supabase / Neon | Managed database (per-project) | App data (client-controlled region) | Client-selected |
| Stripe / Razorpay | Payments (where enabled) | Transaction data | Per provider |
7. Data Retention & Deletion
- Production source & delivered assets: retained for the life of the engagement plus 12 months for support continuity.
- Communications & draft artifacts: retained for 24 months.
- Credentials: revoked or rotated within 7 days of project close.
- On request: clients may request earlier deletion in writing. A written certificate of destruction is provided for regulated engagements.
8. International Transfers & Jurisdiction
SARVAYA is headquartered in Jaipur, Rajasthan, India. Disputes are subject to the exclusive jurisdiction of the courts in Jaipur. For EEA / UK clients, we execute Standard Contractual Clauses (SCCs) as part of our DPA. For US-based clients, we typically operate on the client's MSA with an attached SARVAYA security addendum.
9. Responsible Disclosure
If you believe you have found a security issue in anything SARVAYA owns or operates, email dev.sharma@sarvaya.in with subject line "Security disclosure". We acknowledge within 48 hours and do not pursue good-faith researchers who operate within this policy.
10. Contact & Document Requests
For security questionnaires, DPAs, subprocessor lists, or an AI governance one-pager, reach out:
- Email: dev.sharma@sarvaya.in
- WhatsApp: +91 9371553524
- LinkedIn: linkedin.com/company/sarvaya
Request Our DPA, NDA, or AI Governance Brief
Sent in under 24 hours - pre-reviewed and ready for your legal team.
Request Documents- 5.0 ★★★★★ 25+ Reviews
- Trustpilot ★★★★★ 20+ Reviews
- Clutch ★★★★★ Verified Partner
- NDA-backed Confidential
- 4 Countries IN · US · UK · AU