Copied to clipboard!
ACTIVE ENDPOINT TELEMETRY • ENSEMBLE ML + LLAMA 3.1

AI-Powered, Real-Time Ransomware Detection & Mitigation

Going far beyond signature-based scanning. Ransomware Shield unifies 19-feature Ensemble ML, BLAKE3 Static PE Analysis, Process-Attributed Honeyfile Traps, Live Network Threat Intelligence, and Groq LLaMA 3.1 remediation into a full-stack SOC platform.

Django 4.2 + Channels React 18 + Vite Celery + Redis scikit-learn + YARA Groq LLaMA 3.1
ransomware-shield://soc-command-center
LIVE
Ransomware Shield Platform Banner
BLAKE3: 9f86d081884c7d65... HONEYFILE TRAP: ARMED (3 Decoys)
99.96%
Ensemble ML Precision
14× Faster
BLAKE3 Multi-Hashing
19 Features
Engineered PE Telemetry
60+ APIs
Suspicious Win32 Signatures
< 0.1%
False Negative Rate
INTERACTIVE NORMALIZED SCAN PIPELINE

Live Multi-Engine Threat Simulator

Test Ransomware Shield's weighted scoring formula in real time: VirusTotal (30%) + YARA (25%) + ML Classifier (25%) + Static PE (20%) plus the independent +20% Ransomware String Bonus.

1. Select a Sample PE Payload or Customize Engine Signals Below: Magic Bytes Verified: 4D 5A (MZ/PE32+)

Multi-Engine Signal Weights

VirusTotal Reputation (Weight: 30%) 88 / 100
Thread-Safe YARA Matches (Weight: 25%) 85 / 100
Ensemble ML Confidence (Weight: 25%) 96 / 100
Static PE & Entropy Score (Weight: 20%) 90 / 100
Ransomware String Bonus (+20%)

Triggers on Bitcoin wallets, .onion URLs, YOUR FILES, or vssadmin

Normalized Threat Verdict
CRITICAL Score: 100 / 100
File: invoice_2026.pdf.exe
BLAKE3: e3b0c44298fc1c14...
Extracted Static PE & YARA Indicators
Groq (LLaMA 3.1) AI Threat Analyst Summary Inference: 142ms

MULTI-LAYERED ENDPOINT DEFENSE

7 Integrated Security Engines

Every uploaded binary, file-system event, and outbound socket connection is inspected across seven specialized defense modules.

MODULE 01

Industry-Grade Static PE Analyzer

Inspects Windows Executables without execution, extracting 19 ML-ready features in a single pass with BLAKE3 hashing (~14× faster) plus SHA-256, SHA-1, and MD5.

  • 60+ Suspicious APIs: Crypto (BCryptEncrypt), File Enum, Process Injection (WriteProcessMemory), Anti-Debug, Lateral Movement, & Shadow Copy deletion (vssadmin).
  • Packer & Anomaly Checks: Detects .upx, .aspack, .themida, Authenticode signatures, TLS callbacks, and section entropy.
MODULE 02

Thread-Safe YARA Scanner Engine

High-throughput signature matching built for concurrent Celery worker execution with automated rule hot-reloading.

  • Thread-Safe Caching: Uses threading.Lock with individual file mtime cache invalidation across .yar and .yara rules.
  • Forensic Match Extraction & ReDoS Guard: Extracts first 5 matched strings (× 64 bytes) with strict regex timeouts.
MODULE 03

Ensemble ML Classifier (19 Features)

Soft-voting ensemble combining GradientBoosting + RandomForest trained on 15,550 samples across ransomware, trojans, droppers, worms, and packed malware.

  • Validated Accuracy: Stratified 5-fold cross-validation achieving 99.96% precision and < 0.1% false negative rate.
  • Heuristic Fallback: Multi-signal weighted heuristic guarantees uninterrupted scoring even before model training.
MODULE 04

Generative AI Threat Explainer

Translates dense PE headers, YARA hits, and network telemetry into plain-English incident response playbooks via Groq (LLaMA 3.1) & LangChain.

  • Rich Markdown SOC UI: Renders syntax-highlighted remediation commands, tables, and copy buttons in real time.
  • Conversational Analyst: Interactive security chat assistant plus automated PDF executive report narratives.
MODULE 05

Honeyfile Trap & Process Attribution

Active zero-day ransomware tripwire deploying realistic, non-trivial decoy files (passwords.txt, bitcoin_wallet.dat, tax_returns_2025.pdf).

  • PID & Entropy Forensics: Pinpoints offending process name/path via psutil and flags entropy jumps > 7.0.
  • Instant WebSocket Alerts: Broadcasts CRITICAL alerts over Django Channels with a 10s anti-flood cooldown.
MODULE 06

Live Network Monitor & Threat Intel

Real-time socket telemetry enriched with Emerging Threats IP feeds, MaxMind GeoLite2 country resolution, and ARP Spoofing MitM detection.

  • C2 & Beaconing Detection: Flags C2/mining ports (4444, 8443, 9001, 3333) and detects periodic callbacks via coefficient of variation.
  • 0–100 Connection Risk Score: Maps every socket to its owning PID, executable path, and user account.
MODULE 07 • ORCHESTRATOR CLEAN → LOW → MEDIUM → HIGH → CRITICAL

Normalized Scan Pipeline & Zero-Overwrite Security

Orchestrates VirusTotal (30%), YARA (25%), Ensemble ML (25%), and Static PE Analysis (20%) with an independent +20% ransomware indicator boost. Enforces magic-byte file verification (never trusting extensions), Celery task timeouts (5m hard / 4m soft), and zero-overwrite secure file deletion post-analysis.

CLEAN LOW MEDIUM HIGH CRITICAL
FULL-STACK EVENT-DRIVEN PIPELINE

Interactive System Architecture

Click any architectural layer below to inspect its protocols, asynchronous workers, and underlying Python/JavaScript libraries.

React 18 SOC Frontend Interface
Vite • Tailwind CSS • Zustand • Recharts • React Markdown
PORT 5173
↓ HTTP REST + JWT ↕ WebSockets (ws://)
Django 4.2 API Gateway & Channels (Daphne)
Django REST Framework • SimpleJWT • SQLite3 / PostgreSQL
PORT 8000
↓ Redis Message Broker (localhost:6379) → Celery Background Workers
Parallel Celery Analysis Engines & Groq AI
ASYNC WORKERS
Static PE (19)
YARA Engine
Ensemble ML
VirusTotal + AI
Real-Time Honeyfile Watchdog & Network Threat Intel
watchdog • psutil • geoip2 • Emerging Threats IP Feed
WS STREAM
Component Inspector FRONTEND LAYER

React 18 SOC Frontend Interface

Built with React 18, Vite, and Tailwind CSS to deliver a responsive Security Operations Center dashboard. Uses Zustand for low-overhead global alert state, Recharts for 7-day threat timelines, and React Markdown with syntax highlighting for ChatGPT-style AI threat explanations.

KEY PACKAGES & PROTOCOLS:
UI → POST /api/scanner/upload/ → Polls /jobs/<id>/ & listens on ws/alerts/
LIVE INTERFACE SHOWCASE

Platform Screenshots

Explore the real-time SOC dashboard, live socket risk scoring, Groq LLaMA 3.1 markdown explanations, MitM ARP spoofing alerts, and downloadable PDF reports. Click any screenshot to inspect full-screen.

Dashboard Overview
Expand View

Dashboard Overview

High-level overview of threat scores, clear rates, and recent scans.

Live Network Analysis
Expand View

Live Network Analysis

Real-time connection monitoring with process names, risk scores, and IP reputation.

AI Threat Explainer
Expand View

AI Threat Explainer

LLaMA 3.1 conversational security analyst with ChatGPT-style markdown rendering.

Live MitM ARP Alerts
Expand View

Live MitM ARP Alerts

Global warning for active Man-in-the-Middle ARP spoofing attacks.

Advanced Scanner Engine
Expand View

Advanced Scanner Engine

Multi-engine scan with BLAKE3 hashing, AI explanation, VT results, and ransomware indicators.

Post-Scan PDF Reports
Expand View

Post-Scan PDF Reports

Detailed downloadable reports featuring AI severity explanations.

LOCAL DEPLOYMENT

Quickstart Installation

Requires Python 3.10+, Node.js v18+, Redis (`localhost:6379`), and API keys for VirusTotal & Groq.

bash — Backend & ML Model Training
git clone https://github.com/rudra00434/Ransomware_Shield.git
cd Ransomware_Shield/backend

# Create & activate virtual environment
python -m venv venv
# Windows: venv\Scripts\activate | Mac/Linux: source venv/bin/activate

pip install -r requirements.txt
# Add GROQ_API_KEY & VT_API_KEY to .env

python manage.py migrate
python -m ai_engine.ml.train_model
python manage.py runserver
REST & WEBSOCKET REFERENCE

Interactive API Explorer

Rate Limits: 20 req/min (Anonymous) • 100 req/min (JWT Authenticated).

POST /api/scanner/upload/
multipart/form-data
GET /api/scanner/jobs/<id>/
Poll Status & Results
// GET /api/scanner/jobs/42/ (200 OK)
{
  "id": 42,
  "status": "COMPLETED",
  "file_name": "suspicious_file.exe",
  "result": {
    "threat_level": "HIGH",
    "detection_count": 3,
    "engine_results": {
      "static_analysis": { "entropy": 7.84, "suspicious_apis": 9 },
      "yara_matches": ["Ransomware_LockBit_Strings"],
      "ml_prediction": { "label": "Ransomware", "confidence": 0.994 },
      "virustotal": { "malicious": 48, "total": 72 }
    }
  }
}