intent-based-networking

Intent-Based Networking with Microsoft's Z3 SMT solver. Define intent in YAML, get mathematically proven path diversity and BGP configs deployed to your routers.

An interactive Streamlit web application for network configuration analysis, verification, and differential testing using Batfish.

This project dynamically detects the con-compliant security flow rules based on user Intents.

AGNTCY + LangGraph agents turn intent into CRs; Kubernetes controllers reconcile them onto a live SONiC EVPN/VXLAN fabric; gNMI telemetry closes the loop. Runs in containerlab + kind.

pyATS/Genie pre/post change validator against intent. Fixture CI by default. Lab-only. Not a Mini-SOC.

Architecture for an autonomous control plane over GPU clusters and networks: separated planes, a referee, an autonomy ladder earned by chaos experiments, and a sourced latency hierarchy — from human intent to nanosecond in-ASIC reflexes.

Intent-based network automation platform with full business-to-device lineage. Infrahub SoT + Temporal orchestration + Nokia SR Linux via gNMI. Regulatory audit compliance via data lineage, not archaeology.

Exploring Software Defined Networking concepts, OpenFlow-based network control, and programmable infrastructure using tools like ONOS, Mininet, and Ryu.

Intent-driven policy control for RAN anomaly detection — MSc dissertation code and evaluation artifacts (University of Surrey)

Intent-driven network operations agent — natural-language intent to verified, rollback-safe policy changes. k8sgpt-for-networks.

Benchmark, code and per-seed results for boundary awareness and repair-aware training in SDN intent-conflict detection. Accurate detectors resolve conflicts poorly; verifying explanation faithfulness does not fix it.

Intent Compilation, Multi-Intent Drift Detection, and Adaptive Delegation Envelopes for Autonomous Intent-Based Networking

Official St-INTEL implementation: intent-aware resource allocation with MILP, shadow pricing, Double DQN, and federated learning for 5G Open RAN.

Multi-vendor model-driven network automation framework — NETCONF/gNMI/YANG on Nokia SR Linux & Arista cEOS in a containerlab sandbox.

Routes natural-language network intents to the cheapest LLM that meets a quality floor indexed on the intent's criticality. Semantic complexity estimator (3.4 ms), cost-minimising router, local LLM-as-a-judge. Validated in the data plane on Mininet/Open vSwitch. Master's thesis, CNAM Paris.