MedDevice CyberGuard: AI Agents Producing Cybersecurity Dossiers in Minutes | Amazon Nova Hackathon

MedDevice CyberGuard automates medical device cybersecurity assessment using five specialized AI agents powered by Amazon Nova on AWS Bedrock. What traditionally takes 6 to 12 weeks and costs $50K to $150K in consultant fees is reduced to under 5 minutes at $0.09 per scan.

In this demo, we walk through the full pipeline: conversational device classification (IEC 62304 Class A/B/C), GitHub repository connection, real-time multi-agent analysis, and the final FDA-ready cybersecurity risk dossier.

The system combines real security tools (Semgrep SAST, Syft SBOM) with Amazon Nova 2 Lite reasoning, RAG over actual FDA and IEC regulatory documents via Bedrock Knowledge Bases, and Nova Multimodal Understanding for architecture diagram analysis.

Output includes SAST findings with patient safety context, protocol conformance analysis, STRIDE threat model, SBOM, regulatory compliance matrix (FDA, IEC 62443, IEC 62304, Health Canada), and a prioritized remediation roadmap.

Built for the Amazon Nova Hackathon 2025.

#AmazonNova #AWSBedrock #MedicalDevice #Cybersecurity #GenerativeAI #FDA #IEC62304 #MultiAgentAI #StrandsAgents #Hackathon

Видео MedDevice CyberGuard: AI Agents Producing Cybersecurity Dossiers in Minutes | Amazon Nova Hackathon канала Mohammad Javad Farshchi Heydari