Multi-agent Systems / LLM Orchestration
HERMES BlueTeam.
A local multi-agent architecture for structured event processing and LLM-assisted analysis.
Problem
Organize structured security event processing and analysis within a local platform.
Context
Professional internship project in intelligent application design. Cybersecurity is the application domain; the demonstrated skills are multi-agent AI, LLM orchestration and software architecture.
Objective
Coordinate SOC, SIEM and forensics agents with LLM-assisted analysis.
Dataset / Data
Structured blueteam_event data model and a Suricata demonstration. No dataset volume is claimed.
Architecture
Structured events → specialized agents → LLM analysis → MITRE mapping. Local Docker / Ubuntu environment.
Methodology
Local multi-agent architecture, Suricata demonstration and reported test suite.
My contribution
Designed and developed the platform, orchestrated specialized agents, integrated LLM analysis, processed structured events and carried out functional validation during the internship.
Challenges
Agent coordination, event formats and analysis traceability.
Limitations
Local execution. No commercial deployment, production use at the company or real-world detection rate is claimed.
What I learned
Practice in LLM orchestration, structured processing and containerization.
Project results
Operational agents
Reported local project state; 2 parked skills.
Project tests
Test suite for the local project. No commercial deployment is claimed.