An AI-Powered Cyber Threat Detection & Monitoring System for African Digital Infrastructure .
π SalamaNET AI
AI-Powered Cybersecurity for Africa's Education & Public Sectors
SalamaNET AI is a smart, context-aware cybersecurity system built to detect threats, monitor anomalies, and prevent phishing attacks across public and educational institutions in Africa. Our goal is to enhance digital safety through real-time, automated, and locally-relevant AI solutions.
π¨ Problem Statement
"Increased cyberthreats due to poor cyber hygiene."
Educational and public sector systems in Africa β especially in countries like Kenya β are facing a surge in cyberattacks. Most institutions lack active monitoring tools, suffer from poor password practices, and are unaware of phishing tactics.
π Why It Matters
Kenyan examples:
Public agency data breaches
Schools disrupted by ransomware
Leaked citizen data due to weak security practices
These events reflect a growing crisis that demands an affordable, intelligent, and homegrown solution.
π€ The Solution: SalamaNET AI
SalamaNET AI is an AI-based platform designed to:
π‘οΈ Detect phishing attempts in real time
π‘ Learn continuously from user feedback
βοΈ Empower institutions with actionable visibility into threats
βοΈ How It Works (MVP / Phase 1)
Right now, the MVP (Minimum Viable Product) focuses on phishing detection and learning:
User enters message β Model predicts "safe" or "phishing" + confidence
β User feedback (correct/incorrect)
β Feedback logged & used for retraining
Key Features:
β
Message classification (phishing/safe + confidence score)
β
User feedback system (mark predictions as correct/incorrect)
β
Retraining loop (model improves once enough new feedback is collected)
β
Admin log viewer (daily summaries, filter by model version, view history)
π Project Structure
salamanet_ai/
βββ app.py # Flask web app
βββ retrain_if_needed.py # CLI helper for retraining
βββ train_sklearn.py # Training script (TF-IDF + Logistic Regression)
βββ predictions_log.csv # Saved predictions
βββ feed β¦