FinShield

AI that watches your Application 24/7 - detects breaches, calls your phone, and files the compliance report. Automatically

Comment

Inspiration I kept thinking about the same problem: a fintech startup processing millions of dollars daily and they have to many things and tools to work on. One breach means bankruptcy. One missed compliance deadline means a $500k fine. I wanted to build something that gives a solo engineer the power of an entire security operations center. What it does FinShield monitors your Kubernetes cluster 24/7 and responds to threats automatically. When it detects something suspicious it doesn't send an email it calls your phone. It collects forensic evidence, writes the incident report, maps findings to PCI DSS and HIPAA, and tells you exactly what to do. The AI makes real decisions and predicts which pods will be breached before it happens. How we built it FastAPI backend with PostgreSQL, an autonomous Claude AI loop analyzing threats every 60 seconds, ML models trained on real NVD CVE data, a Go agent running as a Kubernetes DaemonSet, Twilio for voice calls, React frontend, all deployed on Azure Container Apps. Challenges we ran into Getting Claude to return consistent JSON under load, building the DFIR evidence pipeline fast enough to be useful, and making the ML breach prediction actually meaningful without months of real cluster data. Accomplishments that we're proud of The phone call moment. When the attack fires and the phone actually rings with an AI reading the incident report that was the moment it felt real. Also getting 7 compliance frameworks mapped automatically with zero manual configuration. What we learned Security tooling doesn't have to be complex to be powerful. The most impactful features were the simplest ones a real phone call beats a Slack notification every time. What's next for FinShield Real Trivy and kube-bench integration for live scanning, a mobile app for approvals, multi-cluster support, and pricing that actually works for startups because the teams that need this most are the ones who can't afford enterprise security tools.

Built With

Share this project:

Updates