MD. FAHIM HOSSEN
PDFContact

SENIOR AI ENGINEER · AGENTIC SYSTEMS & BACKEND

MD. Fahim Hossen

I design and build production software across AI agents, backend systems, integrations, and modern web applications.

fahimmaria155@gmail.com · github.com/er0r · linkedin.com/in/md-fahim-hossen-47091b15b

View résumé ↗Email me ↗
Current role
Senior AI Engineer
Company
QNE Software Malaysia Sdn Bhd
Focus
Agentic & backend systems
Portrait of MD. Fahim Hossen

Building useful software.
From idea to production.

TECHNICAL TOOLKIT
C# / .NETGoPythonTypeScriptNode.jsReactNext.jsLangChainSemantic KernelPostgreSQLRedisAzureAWSDockerMCP
EXPERIENCE

Professional experience

Senior AI Engineer

QNE Software Malaysia Sdn Bhd

Contributing to QNE’s agentic orchestration platform, QuinnyAI, which works as a Virtual CFO for accounting workflows. Building agent workflows, .NET backend services, MCP integrations, and WhatsApp and Telegram channels.

Lead Software Engineer - Solution Architect and Backend Developer

Flipbizz Innovations Malaysia Sdn Bhd

Architected software systems and database designs, led backend service and API development, and collaborated across product teams. Mentored junior engineers and conducted code reviews to maintain code quality.

Software Engineer

Cleinsight

Focused mainly on backend development, building REST APIs with Laravel, Node.js, and NestJS, alongside infrastructure work using AWS, Kubernetes, Docker, and serverless technologies.

Junior Software Engineer

SJ Innovation LLC

Contributed to client projects using Laravel, PHP, MySQL, WordPress, and MERN-based applications.

EDUCATION

Academic background

Master of Computer Science

Universiti Putra Malaysia

Course-based programme · CGPA 3.75 · Machine Learning · Research Methodology · Research topic: Detection of Disease ↗

Bachelor of Computer Science

American International University-Bangladesh

CGPA 3.68 · Machine Learning · Operating Systems · Object-Oriented Analysis & Design · Data Structures · Algorithms · Database Systems · Software Engineering
SELECTED WORK

Projects

01Featured projectDeveloper tooling

Multiharness

An MCP server, harness-bridge, that lets one coding agent hand work to another agent CLI and read back what happened. Claude Code, Codex CLI, OpenCode or Command Code plans and reviews from the session you already work in, and any of the five agents, Muse Code included, writes the code. Long implementations run as background jobs that outlast agent tool timeouts, every report carries independent Git status and diff evidence, and review findings go back for fixes, up to three rounds. One install line, no Go required, no orchestrator or container.

  • GoBuilds harness-bridge as one small program, installed from a checksum-verified prebuilt binary.
  • MCPRegisters once with the planner and gives it delegate, wait, cancel, route and agents tools.
  • GitAdds git status and diff stats to every report so reviews don't rely on the agent's own account.
Compatible Agents
  1. Plan
  2. Delegate
  3. Report
  4. Review
  5. Repair
Website ↗GitHub repository ↗
02

Closiq

An AI lead-handling product for repeated customer questions and conversations across business channels.

  • LangChainOrchestrates the conversational agent and tool workflow.
  • QdrantRetrieves relevant product knowledge for grounded responses.
  • MinIOStores uploaded knowledge assets and supporting files.
  • DiscordProvides the customer-facing chat interface through a bot.
Website ↗Discord bot ↗
03

Clipinvoice

An invoice management backend for creating, organizing, and tracking invoices through a structured API.

  • NestJSStructures the modular API, validation, and application services.
  • TypeORMMaps entities and manages database access and migrations.
  • PostgreSQLStores invoice, customer, and related transactional data.
  • SwaggerDocuments and exposes the backend API contract.
GitHub repository ↗
04

Detection of Disease

A precision-agriculture project that detects plant diseases from images and connects the prediction to an AI-assisted risk-mitigation advisory workflow.

  • MobileNetV2Classifies rice leaf images into disease categories.
  • Custom CNNProvides a dedicated image-classification pipeline for jute diseases.
  • LLM advisoryTransforms detected conditions into understandable mitigation guidance.
  • MCPConnects disease detection results with the advisory workflow.
Website ↗GitHub repository ↗
05

Azure Project Log

A full-stack workspace that traces Azure DevOps contributions from work items and pull requests through builds and deployment evidence.

  • ExpressRuns the TypeScript modular-monolith backend and secure Azure integrations.
  • ReactProvides the searchable project, task, and contribution workspace.
  • LangGraphCreates an agentic tool-calling loop that selects typed functions, executes them against live Azure data, observes the results, and returns grounded answers.
Website ↗GitHub repository ↗