PII Guard is an LLM-powered tool that detects and manages Personally Identifiable Information (PII) in logs — designed to support data privacy and GDPR compliance.

⚠️ This is a personal side project
Built to explore how Large Language Models can detect sensitive data in logs more intelligently than traditional regex-based approaches.

• About
• Why Use LLMs for PII Detection?
• PII Types Detected
  • Identity Information
  • Sensitive Categories (GDPR Art 9)
  • Government & Financial Identifiers
  • Network & Device Information
  • Vehicle Information
• Architecture
• Getting Started
• Try It Out
• How to Test
• Project Structure
• Suggestions & Contributions

This project experiments with Large Language Models (LLMs) — specifically the gemma:3b model running locally via Ollama — to evaluate how effectively they can identify PII in both structured and unstructured log data.

🧠 LLM-Based Detection with Ollama

• Uses gemma:3b through the Ollama runtime
• Analyzes logs using natural language understanding
• Handles real-world, messy logs better than regex
• Work in progress — contributions welcome!

• 🔍 Identifies PII even when it's obfuscated, incomplete, or embedded in text
• 🌐 Handles multilingual input and inconsistent formats
• 🧠 Leverages semantic context instead of relying on static patterns
• 🧪 Ideal for experimenting with privacy tooling powered by AI

Traditional detection rules often break under complexity — LLMs provide contextual intelligence.

full-name, first-name, last-name, username, email, phone-number, mobile, address, postal-code, location

racial-or-ethnic-origin, political-opinion, religious-belief, philosophical-belief, trade-union-membership, genetic-data, biometric-data, health-data, sex-life, sexual-orientation

national-id, passport-number, driving-license-number, ssn, vat-number, credit-card, iban, bank-account

ip-address, ip-addresses, mac-address, imei, device-id, device-metadata, browser-fingerprint, cookie-id, location-coordinates

license-plate

This is how PII Guard works:

• Clone the repo and start everything with a single command:

make all-in-up

• Shut down everything with:

make all-in-down

This will launch the full stack:

• 🐘 PostgreSQL
• 🔎 Elasticsearch
• 🐇 RabbitMQ
• 🤖 Ollama (with gemma:3b)
• 🌐 PII Guard dashboard and backend API

Visit: http://localhost:3000

http://localhost:8888/api/jobs

curl --location 'http://localhost:8888/api/jobs/flush' \
--header 'Content-Type: application/json' \
--data-raw '{
  "version": "1.0.0",
  "logs": [
    "{\"timestamp\":\"2025-04-21T15:02:10Z\",\"service\":\"auth-service\",\"level\":\"INFO\",\"event\":\"user_login\",\"requestId\":\"1a9c7e21\",\"user\":{\"id\":\"u9001001\",\"name\":\"Leila Park\",\"email\":\"[email protected]\"},\"srcIp\":\"198.51.100.15\"}",
    "{\"timestamp\":\"2025-04-21T15:02:12Z\",\"service\":\"cache-service\",\"level\":\"DEBUG\",\"event\":\"cache_miss\",\"requestId\":\"82c5cc9f\",\"cacheKey\":\"product_44291_variant_blue\",\"region\":\"us-east-1\"}"
  ]
}'

Please refer to the Testing PII Guard guide for instructions on running the test setup, including simulated log generation and stress testing.

This guide will help you set up a test environment to evaluate the performance and detection accuracy of PII Guard.

• API: api/
• Dashboard: ui/
• LLM Prompt Template: api/src/prompt/pii.prompt.ts

Got a bug to report? Feature request? Wild idea? Bring it on!

• 🐛 Bug reports help improve stability
• ✨ Feature requests help shape the product
• 💬 Suggestions, feedback, and contributions are all welcome!