NVIDIA AI Integrations

Gideon is built to leverage NVIDIA AI technologies for high-performance, real-time cybersecurity operations.

NVIDIA Morpheus

NVIDIA Morpheus is an AI framework that provides GPU-accelerated data processing for cybersecurity.

Performance

By using Morpheus, Gideon can process over 208,333 logs per second, allowing for enterprise-scale real-time monitoring.

Enabled Pipelines

Gideon integrates with several Morpheus pipelines:

1. Digital Fingerprinting (DFP): Detects anomalous user behavior by creating a “fingerprint” of normal activity and flagging deviations.
2. DGA Detection: Uses Deep Learning to identify Domain Generation Algorithms used by malware for Command & Control (C2).
3. Phishing Detection: Employs NLP models to analyze email content for social engineering attempts.
4. Ransomware Detection: Monitors file system behavior for encryption patterns and bulk file modifications.

Configuration

MORPHEUS_URL=http://localhost:8080
MORPHEUS_ENABLED=true

NVIDIA NIM

NVIDIA NIM (Inference Microservices) provides optimized, self-hosted AI models.

Gideon can use NIM to host its internal “Self-Reflection” and “Task Planning” LLMs locally on your own GPUs.

Key Benefits

• Privacy: Security research data never leaves your infrastructure.
• Speed: GPU-optimized inference for complex planning tasks.
• Compatibility: Gideon uses an OpenAI-compatible API to talk to NIM.

Configuration

NIM_BASE_URL=http://localhost:8000/v1
NIM_API_KEY=optional

Architecture Diagram

graph TD
    A[Gideon Core] --> B[NVIDIA Morpheus]
    A --> C[NVIDIA NIM]
    B --> D[GPU Acceleration]
    C --> D
    B -- Pipelines --> E[DFP/DGA/Ransomware]
    C -- Local LLM --> F[Mistral/Llama3/Nemotron]