Inspiration
With cyberattacks becoming more common, I realized how easily people can fall victim to online threats. That inspired me to create VigilAI, an AI-powered tool that makes cybersecurity simpler and smarter. It automatically detects and analyzes suspicious online activity in real time. The aim is to help users stay safe without needing deep technical knowledge.
What it does
VigilAI works like a smart cybersecurity assistant that keeps an eye on your online activity in real time. It uses machine learning to spot things like phishing links, unsafe websites, or unusual network behavior. Whenever it detects something risky, it immediately alerts the user and explains what might be wrong. The goal is to help people stay safe online without needing deep technical knowledge.
How we built it
VigilAI combines machine learning and network monitoring to detect online threats. I built the backend in Python using scikit-learn and pandas, and applied NLP to spot phishing patterns in URLs and emails. A Flask dashboard shows alerts in real time. The system flags unusual activity using a simple probability-based anomaly detection model:
Challenges we ran into
Finding good, labeled datasets for cyber threats was tough. The model initially overfitted, so I had to spend time tuning it and selecting the right features. It was also tricky to show AI predictions on a dashboard without slowing the system down. Finally, balancing false alarms with accurate threat detection was a challenge.
Accomplishments that we're proud of
Built an AI-driven cybersecurity assistant from scratch. Detected phishing links and suspicious activity in real time. Created a user-friendly dashboard to display alerts. Applied machine learning and NLP effectively while improving model accuracy
What we learned
How to apply machine learning and NLP to real-world cybersecurity problems. The importance of data cleaning, feature selection, and model tuning for accurate predictions. How to balance system performance and user experience in a real-time application. Practical experience in building a full-stack AI project from backend to dashboard.
What's next for VigilAI -cyber-defense
We want to make VigilAI even more helpful by turning it into a browser extension or API that anyone can use easily.The plan is to use advanced AI to catch more sophisticated cyber threats. We also hope to let users report suspicious activity to help the system learn and improve. Finally, we want to make the dashboard faster and easier to use, so staying safe online is simple for everyone.
Built With
- data
- deployed
- flask
- for
- libraries;
- local/cloud
- mysql
- natural-language-processing
- numpy
- on
- pandas
- python
- scikit-learn
- storage;
