in progress · Contributor · March 2025
AI-Based Vulnerability Detection System
Combines static source-code analysis with PCAP-based behavioral analysis. I contributed to the Python/Django analysis backend and the React/TypeScript interface for uploading inputs and reviewing results.
- Python
- Django
- React
- TypeScript
- Ollama
- Cybersecurity
- Static Analysis
- Behavioral Analysis
- Machine Learning
Built with the PICT Cyber Cell, this prototype combines two forms of analysis to surface vulnerabilities that a single detection method might miss.
Two analysis paths
- Static code analysis accepts a compressed codebase for analysis without executing it.
- Behavioral analysis combines a related codebase with PCAP network-capture data to examine runtime and network behavior.
- A combined workflow brings both sources together to support a more complete vulnerability report.
Engineering approach
- Developed an AI-oriented backend with Django and Python for vulnerability-analysis workflows.
- Worked with the CICIDS2017 network-intrusion dataset and preprocessing notebooks for behavioral analysis.
- Used trained CNN and Random Forest model artifacts as part of the network-analysis pipeline.
- Built a React and TypeScript interface for uploading code and PCAP inputs and presenting analysis results.
The project is collaborative and remains an evolving prototype rather than a production security scanner.