NetGuardXAI NIDS
Industry
Cyber Security and AI
Client
Research Project
Service
Security and Research
Date
April 2025
• Developed and tested a custom CNN-BiLSTM model to detect intrusions on 2M+ network traffic entries; built a deep learning-based IDS achieving 97.03% accuracy through extensive EDA (class balancing, outlier removal, Visualization) and feature engineering (encoding, dimensionality reduction).Combined and preprocessed multi-year CICDS datasets (2017–2020), performed testing and debugging of modules.
• Built a scalable API using Render and AWS Lambda; while also integrated llm API’s likeGoogle Gemini and XAI techniques including LIME and Integrated Gradients for interpretability
A publication for this research findings has been submitted to s2is Journal for review had hopefully will be published soon
For detailed deepdive into this, check out the github repo
Repo - Github
Deployment - Vercel Deployment
