Real company information, real platform metrics pulled live from our database, and a transparent look at how we handle your data and SLAs — nothing fabricated, nothing inflated.
Legal entity details, on the record. No anonymous shell — a registered Indian private limited company.
CYBERDUDEBIVASH PRIVATE LIMITED
CIN: U74999OR2024PTC049281
GST: 21ARKPN8270G1ZP
Founded 2024 · Odisha, India
Bivash — Founder & Principal Security Architect
Cybersecurity practitioner specializing in AI-powered threat intelligence, domain security analysis, and enterprise security architecture.
LinkedIn Profile →PAN: ARKPN8270G
MSME Udyam: UDYAM-OD-19-0133456 (NIC 63122)
Startup India (DPIIT): IN-0426-9439SC
Digital Identity: eMudhra Verified Profile
How scans, AI analysis, and data handling actually work under the hood.
Deterministic, static analysis — not live network reconnaissance. Domain risk scoring runs entirely from domain-string characteristics (TLD reputation, phishing-pattern matching, structural signals); TLS, DNSSEC, HTTP header, port, and email-authentication findings are explicitly flagged as requiring live verification rather than presented as observed. Every scan response carries a machine-readable assessment_mode and live_verification field so this is checkable, not just claimed. See Verified Detection Methodology below.
Scan scoring and threat correlation are deterministic. Conversational AI features (copilot, AI verdicts, executive summaries) are processed by external LLM providers disclosed in our Sub-Processor List, with provider selection and fallback controlled by the platform.
Scan targets and results are stored under your account for history and reports, encrypted at rest, and permanently erased on account deletion (GDPR/DPDP right to erasure).
Honestly scoped — we mark frameworks "In Progress" or "Planning" rather than implying certifications we don't yet hold.
Most AI security vendors report a single "detection accuracy" percentage with no independent audit behind it. Our domain scan engine isn't an ML classifier making claims like that in the first place — it's deterministic, documented business logic, so instead of an unverifiable accuracy number we publish something checkable: every documented risk rule, tested in isolation, with the real pass/fail result from the last CI run.
Loading verified results…
Methodology and full rule list: scripts/domain-engine-rule-conformance.mjs in the public repository. Re-run it yourself against workers/src/engine.js — same input, same result, every time.
Found a security issue? Here's how to report it responsibly.
security@cyberdudebivash.com
Critical: 24h, High: 72h, Medium: 7 days
Private program — contact for access
What you can hold us to.
Instant (automated) to 4 hours (human review)
72 hours standard · 48h premium · 24h enterprise
< 2 hours from NVD publication
< 4 business hours (email), immediate (WhatsApp)
Five steps, start to finish.
Instant Razorpay checkout. GST-compliant invoice sent immediately.
We send a 5-question form to understand your scope and priorities.
Our analyst runs a comprehensive assessment — automated plus manual review.
Full PDF report delivered within 72 hours to your email.
30-minute video walkthrough of all findings with Q&A.
Cloudflare Workers + D1 (SQLite) + KV Storage. Edge-deployed globally with India-region data preference.
JWT authentication, rate limiting, OWASP input validation, and audit logging across all API endpoints.
Threat intel feeds are partially open-source: github.com/cyberdudebivash/cyberdudebivash-ai-security-hub
Third parties that may process data on our behalf. Enterprise customers receive 30 days advance notice before a new sub-processor is added. Questions: privacy@cyberdudebivash.in
Conversational AI features (AI Copilot, AI verdicts, executive summaries) send prompt content — which may include chat messages, scan targets, and scan findings — to one inference provider per request, selected by our provider router with automatic fallback: Groq (US), Cloudflare Workers AI (US/edge), and — only where configured — DeepSeek (China), OpenRouter (US), Together AI (US), Anthropic (US). Deterministic features (scan scoring, template-based rule generation, CVE ingestion) do not call LLM providers.
Talk directly to our security team — no sales script.