Скачать или смотреть Secure Shield | Bot Blocker | SIH- 2025 | SIH25172

IdeaSummary

Web applications are increasingly targeted by complex attacks that traditional rule-based WAFs may miss. This project leverages DistilBERT, a pre-trained transformer, to build an AI-driven WAF capable of understanding patterns in HTTP requests beyond static rules.
The system tokenizes each incoming request, processes it through DistilBERT, and outputs a probability score indicating whether the request is malicious. Requests exceeding a set threshold are blocked, while safe requests are forwarded to the backend.
To ensure the WAF remains up-to-date, the project incorporates a precompiled dataset of known attacks and integrates community and Nuclei templates for CVEs. This enables the model to continuously learn new attack patterns. Docker-based deployment and API endpoints allow easy integration with any website, providing a robust, AI-powered defense layer for modern web applications.