Building an AI Threat Scanner @ Deriv AI Hackathon

Last weekend, my team and I participated in Deriv’s AI Hackathon, where we built an AI-powered threat scanner aimed at making web applications easier to secure.

The idea behind the project was straightforward: security scan results are often difficult to interpret, especially for developers without deep cybersecurity expertise. Our solution acts like an AI security assistant that scans a web application from a provided URL, identifies potential vulnerabilities, and explains risks in clear, human-readable language. Developers can then chat with the AI to better understand findings and receive actionable guidance on how to fix issues.

Within just 24 hours, we delivered a working web application prototype - a challenging but rewarding experience, especially since cybersecurity was a relatively new domain for me. The time constraint forced rapid decision-making, collaboration, and practical problem-solving.

Based on mentor feedback, I’m interested in exploring the next iteration of this project: applying autonomous AI to dynamically adjust scanning strategies in real time based on discovered threats.

This hackathon was a great opportunity to experiment at the intersection of AI, software engineering, and security, while learning how quickly ideas can evolve under pressure.