Portfolio

My Projects

Real network automation and AI tooling built from scratch — everything here runs live and solves problems I have faced working in network operations.

01
Network Automation Lab — AIOps, LLDP topology validation, and switch replacement automation
December 2025
13
network nodes
19
cables validated
8
BGP ASes
15m
AI check cycle

A full enterprise spine-leaf data center built in Containerlab on a free Oracle Cloud VM. The lab runs real eBGP routing with BIRD2, has NetBox as its source of truth, validates its own physical topology using LLDP, automates switch replacement config generation, detects BGP anomalies using Gemini AI via a Flask API, and deploys config changes automatically through GitHub Actions CI/CD. The n8n workflow ties it all together, running every 15 minutes and sending email alerts when anything goes wrong.

How this helps network engineers and teams

This lab demonstrates the full AIOps lifecycle: automated monitoring, AI-driven root cause analysis, human-in-the-loop approval, and automated remediation. For a NOC team, this replaces manual log reviews with a pipeline that catches BGP failures and topology mismatches 24/7. For operations managers, it shows how Python automation reduces the risk of human error during hardware refresh cycles. For individual engineers, LLDP validation and switch replacement automation remove the most error-prone manual steps from a data center migration project.

Python 3.12ContainerlabBIRD2 BGPeBGP AS per rackNetBox v4.5pynetboxlldpdFlask + Gunicornn8nGemini 2.5 FlashGitHub ActionsOracle Cloud ARMNginx + SSL
Read full project
02
Log Redactor — Network log sanitization tool
2025

A browser-based automation tool that strips sensitive data from network logs before they are shared with vendors, TAC teams, or external parties. Built with JavaScript regex patterns, it detects and replaces IPv4 addresses, IPv6 addresses, MAC addresses, BGP community strings, OSPF keys, credentials, and device hostnames. Tokenization ensures the same IP always maps to the same placeholder so the log stays readable and traceable. No data leaves the browser and nothing is stored.

How this helps network engineers and teams

In a NOC or TAC escalation, an engineer typically needs to manually review a log, identify every sensitive field, and scrub it before sending. This takes 10 to 15 minutes per escalation and introduces risk that something gets missed. The Log Redactor does this in under a second with consistent rules across the whole team. For organizations with compliance requirements around customer data and vendor access, it creates a repeatable, auditable process that does not slow down incident response.

JavaScriptRegex tokenizationBrowser-basedNo installNo data storedIPv4 / IPv6MAC addressesBGP communities
Read full project Launch tool
03
Anomaly Detector — AI-powered network log analysis
2025

A web tool that sends network logs to Gemini AI for analysis and returns a structured breakdown of what is wrong, why it is happening, and exactly what to do about it. Engineers paste raw log output from routers, switches, firewalls, or monitoring systems and receive a severity rating (LOW / MEDIUM / HIGH / CRITICAL), a specific list of issues found, a root cause assessment, and step-by-step recommended actions. No account, no install, results in seconds.

How this helps network engineers and teams

During an active incident, a junior or mid-level engineer can paste a 500-line BGP log into the anomaly detector and immediately get a structured triage that would normally require a senior engineer's pattern recognition to produce. For a team of 10 NOC engineers, this means faster escalation decisions, more consistent incident documentation, and less dependence on institutional knowledge held by a single person. For network managers, it means engineers can handle more complex incidents independently, reducing escalation volume to senior staff.

Gemini AIBGP analysisDDoS detectionInterface errorsEVPN / VXLANSecurity threatsSeverity ratingRoot cause analysis
Read full project Launch tool