Security ResearchResearch Prototype

HTTP Anomaly Detection System

Prototype web security research tool

PythonTensorFlowScikit-learn

Overview

An academic research project combining rule-based detection with machine learning to identify and flag malicious web traffic in real time, before it reaches application logic.

Problem Statement

Traditional security tools only catch threats they have seen before. This prototype combines two approaches to catch both known attacks and new, unusual patterns.

Architecture Overview

// SYSTEM_DESIGN

Two-stage pipeline: rule-based filters for known attack types, then a trained machine learning model for behavioural analysis of anything that slips through.

Impact & Outcomes

Achieved 94.7% detection accuracy on a standard benchmark dataset, outperforming signature-only approaches by 18 percentage points.

Next Step

View Details

Back to All Projects