N.I.A.D

Inspiration

UTSA’s strong cybersecurity culture motivated us to explore the future of cyber defense — where AI meets network security. We wanted to build something that shows how even students can contribute to research and innovation in this field.

What it does

N.I.A.D (Network Intrusion & Anomaly Detector) identifies unusual and potentially harmful network behavior. It monitors live traffic and flags anomalies that may indicate intrusions such as port scanning or denial-of-service attempts — providing early threat detection.

How we built it

We divided and conquered the project with two core components: Frontend We built a responsive web interface using HTML, CSS, and JavaScript — featuring a fun Space Cowboy theme to make cybersecurity more exciting and engaging. Backend Tech Stack: Python — Pandas, NumPy, scikit-learn, Flask, Flask-CORS, Joblib

1. Used the NSL-KDD dataset from Kaggle
2. Preprocessed and standardized the data 3.Trained an Isolation Forest model for anomaly detection 4.Integrated the model using a Flask API 5.Captured live packets with Wireshark and evaluated them in real-time

Challenges we ran into

As freshmen and sophomores, working on a full ML-based security system was far above our starting comfort zone. We struggled with: 1.Flask API integration issues 2.Making frontend ↔ backend communication reliable 3.Understanding ML model performance and tuning But thanks to mentors and workshops, we pushed through obstacles and learned a lot.

Accomplishments that we're proud of

We built a functional AI-driven intrusion detector from scratch! We’re proud of: 1.Training and deploying our own model — achieving ~68% detection accuracy 2.Designing a polished, creative UI 3.Learning so many new tools so quickly Most of all, we’re proud that we turned an ambitious idea into a working prototype.

What we learned

This project accelerated our skills in: 1.Machine learning workflows 2.Network traffic analysis 3.Web development 4.API development and integration 5.Real-time anomaly monitoring We now better understand how AI can enhance cybersecurity defense.

What's next for N.I.A.D

We want to upgrade N.I.A.D into a practical security tool that can operate continuously on real networks. Our roadmap includes: 1.More advanced ML models (e.g. GAN-based detectors, autoencoders) 2.Better accuracy through feature engineering and neural networks 3.Dashboard analytics with attack classifications 4.Scalable cloud deployment for organizations We see N.I.A.D evolving into a tool anyone can use to protect their network from hidden threats.

Built With

• css
• flask
• flask-cors
• git
• github
• html
• javascript
• joblib
• nsl-kdd
• numpy
• pandas
• python
• scikit-learn
• venv
• wireshark