All projects
Felt cover for AutoPatch AI

AutoPatch AI

A container-security platform built for the Huawei Developer Competition Europe 2025.

My role: Two-person team: everything except the Huawei integration.

Problem

Teams on Kubernetes run hundreds of container images, and old or risky ones stay in production because the checks are manual. The Huawei Developer Competition Europe 2025 asked for a solution on Huawei Cloud, so the platform had to work against a real CCE cluster.

Approach

A scanner reads pods and images from Huawei Cloud CCE, and a mock mode behind the same interface keeps demos working without an account. A deterministic engine scores each image from 0 to 100 on factors such as the latest tag, root user, legacy tags and image age. TensorFlow.js services add risk prediction and anomaly detection, and an Express API feeds a Next.js dashboard.

Result

The repo ships a working scanner, risk engine and dashboard with alerts, compliance views, scan history, reports and remediation scripts. It is public on GitHub with unit, integration and performance tests.

Highlights

  • Real Huawei Cloud CCE scanner
  • Risk score 0 to 100 per image
  • Mock mode for demos without a cloud account

Stack

  • TypeScript
  • Node.js
  • Express
  • MongoDB
  • TensorFlow.js
  • Socket.IO
  • Next.js
  • React
  • Recharts
  • Jest
  • Docker
  • Huawei Cloud CCE
AutoPatch AI overview dashboard
Overview: risk score, clusters, images and compliance at a glance (no backend connected).
Alerts and notification channels
Alerts by severity with notification channels for email, webhook, Slack and Teams (sample data).
Compliance screen
Compliance status per framework: PCI-DSS, SOC2 and ISO27001 controls (sample data).
Scan history
Scan history with targets, duration, status and findings (sample data).