CYBER SHIELD

🛡️ Inspiration

With the rapid rise of sophisticated digital scams, traditional static blocklists are no longer enough to protect everyday users. Phishing attempts and malicious links now spread dynamically across messaging platforms and emails, often slipping past standard security filters. We wanted to build a proactive, intelligent solution that acts as a real-time shield, analyzing textual content on the fly to protect users—especially vulnerable groups—right where they communicate.

⚙️ What it does

Cyber Shield is an AI-powered browser extension designed to secure your digital interactions in real time. It scans incoming text on web-based messaging platforms and emails, leveraging advanced Natural Language Processing (NLP) models to instantly detect and flag phishing attempts, scams, and deceptive content before a user can click a malicious link or fall victim to fraud.

🛠️ How we built it

Frontend: Developed as a lightweight Chrome Extension using HTML, CSS, and JavaScript to ensure seamless real-time UI/UX overlays and content scripting. AI/NLP Engine: Utilized advanced language models (fine-tuned BERT/mBERT architectures) optimized for text classification and threat identification. Backend & Database:Integrated with Firebase for fast, secure authentication, real-time data handling, and configuration management.

🚀 Challenges we faced

Real-time Latency:Running NLP inference quickly enough to scan incoming text without degrading the user's browsing speed was a major hurdle. We optimized the model architecture and pipeline to ensure split-second detection. Contextual Accuracy: Distinguishing between safe, casual conversations and actual social engineering tactics required careful data preprocessing and model fine-tuning to keep false positives low.

💡 Accomplishments that we're proud of

• Successfully engineered a working end-to-end pipeline from a browser content script to an NLP decision engine.
• Created a highly accessible, low-overhead security tool that works silently in the background to protect users without requiring complex setup.

🧠 What we learned

• Gained deep, hands-on experience in deploying machine learning models specifically for defensive cybersecurity applications.
• Mastered asynchronous JavaScript and real-time state management within browser extensions.

🔮 What's next for Cyber Shield

• Expanding platform support to edge-case messaging networks.
• Enhancing the detection engine to support multi-lingual threat identification natively.
• Implementing a unified accessibility dashboard to offer tailored safety profiles for different user demographics.

Built With

• bert
• chrome
• css3
• firebase
• html5
• javascript
• natural-language-processing
• python