What it does

Project Siren — Project Story

Inspiration

Cybersecurity systems often generate huge amounts of security events, but detecting a threat is only the first step. The real challenge is deciding what should happen next—whether an event is harmless, suspicious, or dangerous enough to require immediate action.

This inspired me to build Project Siren, an AI-powered cybersecurity gateway designed to act as an intelligent layer between applications and potential threats. I wanted to explore how AI could move cybersecurity from simply detecting problems toward understanding threats and automatically responding to them.

What I Built

Project Siren combines security monitoring, AI-based analysis, vulnerability detection, and automated policy decisions into a single security gateway.

The system can:

• Monitor and log security events in real time.
• Analyze suspicious activity using AI.
• Scan dependencies for known vulnerabilities using the OSV vulnerability database.
• Maintain security policies and threat information.
• Assign threat severity and convert it into an appropriate security decision.
• Automatically ALLOW, WARN, or DENY requests based on the security context.
• Provide a SOC-style dashboard for monitoring security activity.
• Track communication between services through a communication trace.
• Support automated threat response and lockdown mechanisms.

The core idea is simple:

Detect → Analyze → Decide → Respond

How I Built It

I built the project using React, TypeScript, Vite, Tailwind CSS, Supabase, and serverless functions.

The frontend provides the security operations dashboard and dispatcher interface, while Supabase handles data storage, authentication, realtime security events, and backend functions.

For vulnerability intelligence, I integrated the OSV API to identify known vulnerabilities in dependencies. I also experimented with AI-powered semantic interrogation to allow security events to be analyzed beyond simple rule-based detection.

The decision engine maps threat severity to security policies. For example, critical threats can trigger a DENY decision, while lower-risk events can be allowed or flagged for further attention.

I also implemented realtime event monitoring so that newly detected security events can automatically influence the security state of the system.

Challenges I Faced

One of the biggest challenges was integrating multiple components while keeping the system reliable and understandable.

I faced issues with:

• Connecting frontend components to Supabase backend functions.
• Integrating external vulnerability and AI APIs.
• Handling AI model availability and API quota limitations.
• Designing automatic responses without creating unnecessary false positives.
• Implementing realtime security-event monitoring.
• Making sure automated decisions did not continuously switch between states due to temporary events.

To address the last problem, I introduced a rolling time-window approach for threat analysis, helping the system make decisions based on a broader security context rather than reacting excessively to a single event.

What I Learned

Building Siren taught me that cybersecurity is not just about finding vulnerabilities—it is about making reliable decisions from incomplete and constantly changing information.

I learned how to connect a modern React application with backend services, work with realtime data, integrate external security APIs, design security decision logic, and use AI as part of a larger cybersecurity workflow.

Most importantly, I learned that AI is most useful when it works together with deterministic security rules rather than replacing them completely.

What's Next

I want to continue developing Siren into a more autonomous security gateway capable of understanding complex attack patterns, correlating events across services, and recommending or executing appropriate responses with minimal human intervention.

Project Siren is my attempt to build a cybersecurity system that doesn't just sound the alarm—it understands the threat and knows what to do next.

Built With

• css
• react
• supabase
• tailwind
• typescript