home
login
threat detection
chat space end to end
bert algorithm working
private Direct Messages

Inspiration

Defence personnel, veterans, and their families are increasingly targeted by AI-generated phishing, social engineering attacks, impersonation, and disinformation campaigns. While encrypted messaging apps protect data transmission, they do not protect users from intelligent threats inside conversations.

We asked a simple question:

What if AI could defend conversations before they become threats?

That idea led to SentinelNet — an AI-defended secure communication platform designed for defence-grade environments.

What it does

SentinelNet is a secure communication ecosystem that combines:

• End-to-End Encryption (AES-256)
• AI-powered threat detection
• OPSEC (Operational Security) leak detection
• AI-generated content identification
• Phishing & social engineering detection
• HQ Command Monitoring Dashboard

Before a message is encrypted and sent, it is analyzed by multiple AI models to detect:

Sensitive operational information
Suspicious urgency language
Impersonation patterns
AI-generated manipulation

If risk is detected, users are warned in real time.

Admins can monitor anonymized risk insights through a defence-grade HQ dashboard.

SentinelNet doesn’t just encrypt messages — it actively protects users.

How we built it

Frontend:

Next.js + React
Tailwind CSS
Recharts for analytics visualization

Backend:

FastAPI (Python)
WebSockets for real-time communication
PostgreSQL (Supabase-ready)

Security:

AES-256 encryption
Public-private key exchange
JWT authentication
Forward secrecy session keys
SHA-256 message integrity hashing

AI Threat Intelligence Engine: We built a hybrid AI pipeline using HuggingFace Transformers and PyTorch.

Models include:

AI-generated content detector (DistilBERT/RoBERTa fine-tuned)
OPSEC risk classifier
Phishing & social engineering detector

Each message passes through: Feature Extraction → Multi-model inference → Risk Aggregator → Explanation Layer → Encryption

Challenges we ran into

Integrating AI scanning without compromising end-to-end encryption logic.
Designing risk detection that is sensitive enough to detect threats but not overly aggressive.
Handling secure key exchange and forward secrecy implementation.
Managing database authentication and user role-based access.
Designing a UI that feels defence-grade instead of casual messaging style.

Balancing security, usability, and AI performance was our biggest technical challenge.

Accomplishments that we're proud of

• Built a working AI-powered threat scanning pipeline
• Integrated encryption with pre-send AI analysis
• Developed a real-time HQ monitoring dashboard
• Designed a professional defence-style UI
• Successfully implemented role-based governance
• Created a modular architecture for future scalability

The biggest achievement: creating a system that shifts communication security from passive encryption to active AI defense.

What we learned

• Encryption alone is not enough in modern cyber warfare.
• AI can be used defensively to protect communication ecosystems.
• Security systems must balance privacy and intelligence.
• Designing for high-risk environments requires disciplined architecture.
• Real-time AI inference must be optimized carefully for latency.

We learned how to combine cybersecurity, AI, and system design into one unified ecosystem.

What's next for SentinelNet – AI-Defended Secure Communication Platform

Future roadmap includes:

• On-device AI inference for zero server visibility
• Blockchain-based audit logs
• Behavioral anomaly detection
• Federated learning for secure model updates
• Government-grade deployment infrastructure
• Mobile app deployment (Android & iOS)
• Secure file intelligence scanning
• Real-time threat clustering and visualization

Our vision is to transform encrypted messaging into an intelligent defensive communication network.

SentinelNet aims to become the future of secure, AI-protected communication systems.