Noufal Digital LLC — audit.sekurely.io

We Find What's Leaking
Inside Your AI

No compliance PDFs. No generic checklists.
Only real vulnerabilities found using live tools.
✅ Sekurely.io — live AI scanner we built & shipped
✅ 4 tools run hands-on: Presidio · Langfuse · ChromaDB · LiteLLM
✅ Hands-on — NIST-based controls, prompt injection testing, AI threat modeling
✅ 30-Day AI Architecture bootcamp completed
⚡ Limited to 5 audits per week — hands-on work, not automated scans

Who needs an AI Security Audit?

If your business uses AI tools and handles real customer data — you need this.

🛒
E-commerce & Shopify Stores
Using AI for support, descriptions, or order automation? Customer names, emails, and order data are likely flowing through unprotected prompts right now.
⚠ Risk: PII leakage, no data policy
💼
SaaS Founders with AI Features
You built an AI chatbot into your product. Have you tested it for prompt injection? Can users extract your system prompt or manipulate responses?
⚠ Risk: Prompt injection, system prompt leak
🏢
SMBs Using AI Automation
Running n8n, Zapier, or Make with AI? Your API keys and customer data may be flowing through pipelines with zero logging or monitoring.
⚠ Risk: Exposed API keys, no audit trail
🤝
Agencies Building for Clients
You deliver AI solutions but haven't security-tested them. One breach reflects on you. Add our audit as a deliverable and charge more.
⚠ Risk: Client liability, trust damage
🏥
Healthcare & Wellness Apps
Any AI touching patient info must be HIPAA-aware. Most AI tools are not compliant out of the box — and most founders don't know it.
⚠ Risk: HIPAA violation, sensitive data in LLM
📊
Finance & Legal Teams
Using ChatGPT for contracts or modeling? Confidential data doesn't belong in public LLMs without enterprise data agreements in place.
⚠ Risk: Data exposure, compliance gap

Real work. Real proof.

Not certifications. Things we actually built, ran, and shipped — on our own VPS, with real tools.

LIVE PRODUCT
Sekurely.io — AI Security Scanner
Built and shipped a live SaaS with 3 active scanners: website/domain, CVE lookup, and email breach checker. Powered by n8n, OpenAI, Supabase, VirusTotal, and Google Safe Browsing. Running in production today.
→ sekurely.io
BOOTCAMP CAPSTONE
Private Financial Analyst AI
Production RAG system with PII stripping via Presidio, Cohere reranking, Langfuse cost tracking, Ragas evaluation, Prompt Guard, Arize Phoenix monitoring, and Terraform IaC on AWS. Every security layer implemented hands-on.
→ GitHub portfolio
presidio-pii-audit.py — VPS srv1589904 — 2026-04-25
===== PRESIDIO PII AUDIT SCAN =====
Auditor: Noufal Digital LLC

[SAMPLE 1] Customer John Smith emailed john.smith@gmail.com...
  ⚠ DETECTED: EMAIL_ADDRESS | confidence: 1.00
  ⚠ DETECTED: PERSON | confidence: 0.85
  ✓ ANONYMIZED: Customer <PERSON> emailed <EMAIL_ADDRESS>

[SAMPLE 5] User sarah.jones@company.com logged in from IP 192.168.1.10
  ⚠ DETECTED: EMAIL_ADDRESS | confidence: 1.00
  ⚠ DETECTED: IP_ADDRESS | confidence: 0.95
  ✓ ANONYMIZED: User <EMAIL_ADDRESS> logged in from IP <IP_ADDRESS>

===== SCAN COMPLETE =====
langfuse-observability-audit.py — VPS srv1589904 — 2026-04-25
===== LANGFUSE OBSERVABILITY AUDIT =====
Auditor: Noufal Digital LLC
Time: 2026-04-25 07:10:19

✓ Langfuse connection: OK
✓ Observability layer: ACTIVE

===== OBSERVABILITY CHECK COMPLETE =====
chromadb-vector-audit.py — VPS srv1589904 — 2026-04-25
===== CHROMADB VECTOR STORE AUDIT =====
Auditor: Noufal Digital LLC | Time: 2026-04-25 07:15:17

✓ Vector store connected
✓ Collection size: 5 documents

Scanning for PII in vector store chunks:

  [⚠ PII FOUND IN VECTOR STORE]
   Chunk: Customer John Smith SSN 123-45-6789 purchase history...
  [⚠ PII FOUND IN VECTOR STORE]
   Chunk: User email john@gmail.com credit card 4111-1111-1111...
  [⚠ PII FOUND IN VECTOR STORE]
   Chunk: Patient record DOB 01/15/1985 diagnosis: hypertension...

FINDING: Unfiltered PII detected in vector store chunks
FIX: Run Presidio anonymizer before indexing documents

===== CHROMADB AUDIT COMPLETE =====

What your report looks like

Every client gets a structured findings report — severity ratings, tool evidence, and a clear fix roadmap.

AI_Security_Audit_Report_[Client].pdf — SAMPLE
AI Security Audit Report
Client: [E-commerce SaaS]  ·  Auditor: Ammara Noufal, Noufal Digital LLC  ·  Date: April 2026
Tools Used: Presidio · LiteLLM · Langfuse · ChromaDB · Manual Injection Testing
3
CRITICAL
4
HIGH
2
MEDIUM
38/100
RISK SCORE
FINDINGSEVERITYTOOLFIX
Customer email addresses detected in 6 of 10 AI prompts sampledCRITICALPresidioAdd Presidio anonymizer before prompt construction
System prompt extracted via roleplay jailbreak in 2 attemptsCRITICALManual TestHarden system prompt + add injection detection layer
OpenAI API key found in plain text in GitHub repoCRITICALManual ReviewRotate immediately. Move to AWS Secrets Manager
No rate limiting on public AI chatbot — DoS and cost attack possibleHIGHLiteLLMConfigure rate limits via LiteLLM gateway
Zero logging on AI API calls — no way to detect misuseHIGHLangfuseDeploy Langfuse tracing. Set anomaly alerts
PII found unfiltered in ChromaDB vector store chunksMEDIUMChromaDBRun Presidio on all documents before indexing
SAMPLE REPORT — Noufal Digital LLC · audit.sekurely.io · Representative example of audit output

Live Prompt Injection & PII Scanner

Pick an attack pattern or type your own. See what our detection logic catches.

sekurely-scanner — Live Detection Engine
LOAD ATTACK PATTERN:
INPUT TO SCAN:
⚡ Note: Demo shows detection logic. Full audit runs deeper backend scans with Presidio, LLM Guard, and Langfuse on your actual system.

Audit Packages

Flat-rate pricing. No hourly surprises. Delivered in days, not weeks.

⚡ Limited to 5 audits per week — hands-on work, not automated scans
TIER 01 — STARTER
$149 flat
⚡ 3-day delivery
  • Presidio PII scan on your prompts
  • Manual prompt injection test
  • Top 3 critical risks flagged
  • 1-page findings PDF
  • 48hr email support
TIER 02 — STANDARD
$349 flat
⚡ 5-day delivery
  • Full Presidio PII scan
  • Prompt injection + jailbreak test
  • LiteLLM gateway review
  • Langfuse observability check
  • ChromaDB / vector store scan
  • Full findings report (see sample)
  • Remediation roadmap
  • 30-min debrief call
TIER 03 — PREMIUM
$699 flat
⚡ 7-day delivery
  • Everything in Standard
  • AWS IAM + Bedrock audit
  • RAG pipeline security check
  • Arize Phoenix monitoring setup
  • AI usage policy draft
  • 60-min strategy call
  • 30-day follow-up check

Ready to find out what's exposed?

Most businesses using AI have at least one critical vulnerability they don't know about. Share access, I run the scans, you get a clear report in 3–7 days. No fluff.

01
You share access
API keys, prompts, or repo — only what's needed for the scope
▶
02
I run live scans
Presidio, Langfuse, ChromaDB, manual injection tests on your actual system
▶
03
You get the report
Full findings PDF with severity ratings, tool evidence, and fix roadmap — in 3–7 days
▶
04
Debrief call
Optional 30–60 min call to walk through findings and answer questions
→ HIRE ON UPWORK 💬 WHATSAPP ✉ EMAIL
⚡ Limited to 5 audits per week — hands-on work, not automated scans
UPWORK PROPOSAL — COPY & SEND
Hi [Name], I noticed you're building with AI — I run practical security audits using real tools, not checklists. What I actually do: • Presidio scan — detects PII leaking into your LLM prompts • Manual injection test — attempts to extract your system prompt and override instructions • LiteLLM gateway review — rate limits, cost exposure, access control gaps • Langfuse check — are your AI calls even being logged? • ChromaDB / vector store scan — PII sitting unfiltered in your RAG pipeline I built Sekurely.io (live AI security scanner) and completed a 30-day AI architecture bootcamp using these tools hands-on in production. You can see real scan output and a sample report at audit.sekurely.io Quick question: do you have a customer-facing AI chatbot, or is this internal tooling? That determines which scan matters most. Ammara | Noufal Digital LLC audit.sekurely.io
LINKEDIN POST
🔐 We run actual AI security scans. Not checklists. Not PDFs. Here's what we find in almost every audit: ❌ Customer PII flowing through unprotected prompts (caught with Presidio) ❌ System prompts extracted via jailbreak — took under 2 minutes ❌ API keys in GitHub repos — found in 2 of our last 3 reviews ❌ PII sitting unfiltered inside ChromaDB vector stores ❌ Zero logging on AI calls — no way to detect misuse We built Sekurely.io and use Presidio, LiteLLM, Langfuse, and ChromaDB in real audits — not just read about them. See real scan output + sample report: audit.sekurely.io DM me "AUDIT" — first 15-min call is free. #AISecurity #LLMSecurity #Cybersecurity #AIAudit