Skip to content

Featured Projects

Network discovery tooling, log analysis pipelines, a local-first desktop ops platform, an AI-assisted German law research app, and this portfolio — built across Earth Telecommunication, tripunkt GmbH, and personal projects.

network discovery & topology mapping.app
Subnet Sweep Config
ICMP+SNMP Scan
JSON Relationship Build
Interactive Topology Render
Network Operations · Incident Triage

Network Discovery & Topology Mapping

Python-based tool for live host scanning and interactive relationship visualization. Built at Earth Telecommunication (2013–2015).

Context

Earth Telecommunication lacked centralized network documentation. Maps were scattered across spreadsheets and engineer memory, complicating incident resolution for over 500 accounts.

Implementation

Built a Python tool to scan live hosts using ICMP and SNMP sweeps, then aggregated device relationship data into structured JSON. The output rendered as an interactive JS/HTML topology graph enabling real-time component isolation and data-path tracing during incidents. Scanner aggressiveness was tuned per subnet to balance completeness against network load.

Outcome

Enabled real-time component isolation and data-path tracing during network incidents, reducing Mean Time to Resolution (MTTR). Engineers visually isolated faulty segments in seconds rather than tracing cables manually. The tool became the de facto reference for incident triage across the NOC team.

Impact
Became the de facto reference for NOC incident triage
Tech Stack
PythonJavaScriptHTMLTopology MappingNetwork Operations
log analysis & automated ticketing.app
Log Stream Ingestion
Regex Incident Detection
Threshold Evaluation
Jira Ticket Dispatch
Production Observability · Incident Response

Log Analysis & Automated Ticketing

Automated error detection and Jira ticket creation workflow.

Context

Critical error patterns were buried in thousands of log lines per minute. Subtle regressions went unnoticed for hours. Engineering needed real-time detection, not post-mortem discovery.

Implementation

Developed a Python-based log parsing workflow that monitors error frequency against configurable thresholds using regex pattern matching. When a validated incident exceeds its threshold, the system automatically creates a Jira ticket with contextual stack traces and surrounding log context via the Jira REST API. The pipeline runs on a scheduled cron trigger and supports multi-service log sources.

Outcome

Reduced response times for recurring production errors from hours to minutes and improved incident tracking reliability with structured Jira artifacts. Alert fatigue was significantly reduced by tuning thresholds per error pattern rather than using one-size-fits-all rules. The system now handles over 200 log streams daily.

Impact
Handles over 200 log streams daily with per-pattern adaptive thresholds
Tech Stack
PythonRegexJira APIOperational Support
universalops.app
Go Backend Metrics
Wails Native Bindings
React UI Layer
Ollama Local AI
Desktop Operations · Local-First Engineering

UniversalOps

Native local-first desktop operations platform for system monitoring, network diagnostics, and security auditing. Built with Go, Wails v2, and React 19.

Context

Existing monitoring tools required cloud backends, sent telemetry out, or couldn't run offline. Needed a unified operations toolkit that worked entirely on a local machine.

Implementation

Built with Go and Wails v2 for native OS integration, React 19 frontend with Tailwind CSS, and SQLite for local persistence. Integrates Ollama for on-device AI-assisted diagnostics. Features real-time system monitoring, network scanning, container management, and security auditing in a single desktop app.

Outcome

Fully offline operations platform with zero telemetry — system monitoring, network scanning, container management, and security auditing running locally. Local AI via Ollama provides diagnostic assistance without sending data to any third party.

Impact
Fully offline ops platform with zero telemetry and local AI
Tech Stack
GoWailsReactTypeScriptSQLiteOllama
ai-assisted german law.app
Hybrid Search Query
Qdrant Vector Retrieval
AI Guidance Engine
RDG-Compliant Output
Legal Tech · Semantic Search

AI-Assisted German Law

Search, browse, and receive AI-guided analysis of 6,000+ German federal laws with multilingual support and 4 chat modes.

Context

German legal text is dense and scattered. Navigating it without expensive legal counsel is nearly impossible for non-lawyers and non-native speakers.

Implementation

Built with Next.js 16, Supabase (Postgres + RLS), and Qdrant Cloud for vector search. Features hybrid BM25 + dense retrieval across 103K+ vector points, 9-language i18n, AI guidance engine with 4 chat modes (Basic/Browser/Cloud/Local), bookmark system, remediation playbooks, and document templates.

Outcome

302 tests passing, 103K+ indexed law sections in Qdrant, hybrid search with 3-stage fallback chain, AES-256-GCM encrypted API key storage, and full RDG-compliant disclaimers. Deployed on Vercel with 9-language UX.

Impact
302 tests, 103K+ vector points, 9-language AI guidance
Tech Stack
Next.jsTypeScriptSupabaseQdrantPostgreSQLPython
shahriarhaqueabirportfolio.app
Next.js Route Render
Framer Motion Layout
WebLLM Worker Init
Fallback Intent Match
Frontend · AI Integration

ShahriarHaqueAbirPortfolio

Personal portfolio site with in-browser AI, parallax animations, and a SPA-in-MPA architecture on Next.js 16.

Context

Wanted a portfolio that demonstrates both frontend craftsmanship and practical AI integration — not just a static resume site.

Implementation

Built with Next.js 16 (App Router, webpack) and React 19. Features Framer Motion parallax/timeline animations, WebLLM-powered Qwen2.5 1.5B in a web worker for fully local AI chat, a fallback intent engine with 17 patterns, voice I/O, tsParticles backgrounds, and Tailwind CSS v4.

Outcome

13/13 static routes, SPA-like navigation with MPA reliability, local AI running entirely in the browser via WebGPU, and sub-4s production builds.

Impact
Local AI chat in-browser with 13/13 static routes
Tech Stack
Next.jsReactTypeScriptFramer MotionWebLLMTailwind CSS
Suggested Next Paths

What has he built?

Technical implementations, AI systems, AI automation workflows, architecture, and case studies.