Closed‑loop AI network automation — detect, remediate, and verify.
Inspired by multivendor‑lab architecture: RFC 6241 confirmed‑commit rollback, anomaly auto‑remediation, and immutable audit trail.
AI-driven insights for network operations — combining routing data (BGP/OSPF) with large language models for anomaly detection, policy automation, and intelligent troubleshooting.
This proof-of-concept explores how AI can enhance Border Gateway Protocol (BGP) and OSPF operations. By integrating LLM-driven analysis with routing data, the project demonstrates potential applications such as anomaly detection, automated policy suggestions, and intelligent troubleshooting.
🏗️ Architecture https://copilot.microsoft.com/th/id/BCO.767b4425-3e03-4d4f-b506-a69dbb2be4a0.png
+-------------------+ | Data Sources | <-- BGP/OSPF routing data, configs +-------------------+ | v +-------------------+ | RAG Pipeline | <-- Retrieval-Augmented Generation +-------------------+ | v +-------------------+ | LLM Runner | <-- AI-driven analysis & suggestions +-------------------+ | v +-------------------+ | Reports/Tests | <-- OSPF test reports, CI/CD outputs +-------------------+
Key Components llm_runner.py → Core AI execution logic
rag/ → Retrieval-Augmented Generation modules
scripts/ → Utility scripts for automation
tests/ → Unit & integration test cases
jenkins/ → CI/CD pipelines for static analysis
sonar-project.properties → SonarQube integration
Features AI-driven insights for BGP/OSPF routing
Automated test reports (ospf_test_report.txt)
Jenkins pipelines for static analysis
SonarQube integration for code quality
Modular design for extensibility
Testing Run all tests:
pytest tests/
Integration test example: python tests/integration/ospf_test.py
Sample output: ================== test session starts ================== collected 12 items
tests/test_utils.py ..... tests/test_rag.py ..... tests/test_llm_runner.py ...
================== 12 passed in 2.34s ==================
Roadmap Expand unit test coverage
Add real-world BGP datasets
Enhance logging & error handling
Publish demo notebooks with sample outputs
# Clone the repository
git clone https://github.com/imsusovan/ai-network-ops.git
cd ai-network-ops
# Install dependencies
pip install -r requirements.txt
# Run the AI-Network-Ops integration
python llm_runner.py --config config.yaml