AI security portfolio
AI security projects and autonomous agents
I built these projects to explore three different security problems: automatically finding and fixing vulnerable code, coordinating specialized security agents, and protecting users across browser and desktop workflows.
Project comparison
| Project | Primary focus | Approach | Best fit |
|---|---|---|---|
| Sast.tech | Automated web security testing and code remediation | One autonomous agent finds vulnerabilities, verifies them, patches code, and reports the result. | Developers who want security checks inside a fast software delivery workflow. |
| SWRMZ | Cooperative AI agents for application security | A swarm of specialized agents hunts vulnerabilities, remediates issues, and guards runtime logs. | Teams exploring multi-agent coordination across testing, remediation, and monitoring. |
| Zero Threat | A broader AI-powered cybersecurity ecosystem | A web platform, browser extension, and Windows agent combine download checks and malware analysis. | Users who need security coverage across browsing, files, and endpoint workflows. |
What these projects taught me
Security automation is only useful when findings are reproducible. That is why the Sast.tech workflow emphasizes proof-of-concept testing and verification after a patch, rather than treating every scanner alert as equally reliable.
Agent specialization can make a complex workflow easier to reason about. SWRMZ separates discovery, remediation, and runtime monitoring so each agent can focus on a narrower task while still contributing to one security objective.
Security also extends beyond source code. Zero Threat combines web, browser, and Windows components because risky downloads and endpoint behavior require a different layer of protection than application testing.
Frequently asked questions
What is an autonomous AI security agent?
It is a software agent that can perform a security workflow with limited manual intervention, such as discovering an application, testing for weaknesses, validating findings, and proposing or applying fixes.
How is Sast.tech different from a static code scanner?
Sast.tech is designed around an end-to-end workflow: it inspects a live application and repository, tests findings, patches vulnerable code, and verifies the result instead of stopping at a list of warnings.
Why use multiple AI agents for cybersecurity?
A multi-agent design can separate responsibilities such as reconnaissance, vulnerability analysis, remediation, and log monitoring while allowing the agents to share context and coordinate decisions.