Full-Stack Engineer · AI/ML

I build full-stack products with AI/ML woven into the architecture, not bolted on as a feature.

I design and build full-stack SaaS products — from multi-tenant architecture and secure data isolation to the dashboards and workflows that run on top of them. My work blends production engineering with applied AI/ML, using LLMs, RAG pipelines, and automation to make software smarter without adding unnecessary complexity. I care about systems that hold up under real use: secure by default, built to scale, and shipped without shortcuts.

View my workGet in touch
Muhammad Muaaz Ulhaq

ABOUT

How I work

I design and build full-stack SaaS products — from multi-tenant architecture and secure data isolation to the dashboards and workflows that run on top of them. My work blends production engineering with applied AI/ML, using LLMs, RAG pipelines, and automation to make software smarter without adding unnecessary complexity. I care about systems that hold up under real use: secure by default, built to scale, and shipped without shortcuts.

Punjab, Pakistan

FEATURED PROJECTS

Selected work

FLAGSHIP PROJECT

DealerHelm

View live app

A multi-tenant SaaS platform for car dealership inventory management, replacing the spreadsheet-based workflows most independent dealerships still rely on.

Problem

Independent dealerships run inventory, financials, and operations out of Google Sheets — no real-time visibility, no data isolation, no room to scale.

What I built

A multi-tenant SaaS platform with centralized inventory management (vehicle records, VIN-based search, stock tracking), real-time operational dashboards, and financial tracking tied directly to inventory (costs, sales, margins). Every dealership's data is fully isolated from every other tenant on the platform.

  • Backend enforces tenant isolation using JWT claims — every request is scoped by the sub claim, so tenant boundaries hold even under claim-tampering attempts
  • Built and tested against real attack surfaces for multi-tenant SaaS: cross-tenant access via staff roles, sub-resource isolation, race conditions
  • Structured, professional-grade codebase — logic separated across files rather than monolithic scripts, built with an agentic dev workflow (VS Code + Claude Code)

Sole/lead developer — architecture, backend, multi-tenancy/security design, and the frontend dashboard build.

Next.jsTypeScriptPostgreSQLJWTMulti-tenant architecture

In active development

AI-Powered Weather Forecast & Visualization App

An AI-driven weather app delivering forecasts, interactive charts, historical tracking, and AI-generated recommendations.

Problem

Raw weather data is hard to act on — most weather apps show numbers without context or guidance.

What I built

Built the full pipeline — backend API, data visualization layer, and a Gemini-powered recommendation engine that turns forecast data into plain-language, actionable guidance.

  • FastAPI backend serving forecast and historical data
  • Streamlit frontend with interactive charts and historical tracking
  • SQL for persistence and history
  • Gemini API integration for AI-generated recommendations

Sole developer — backend, frontend, and AI integration.

FastAPIStreamlitSQLGemini API

Shipped

Information Flow–Guided Grey-Box Fuzzing for Smart Contract Systems

A final year research project proposing a novel information-flow-guided fuzzing technique to improve smart contract security testing with lower resource overhead than standard grey-box fuzzing.

Problem

Standard grey-box fuzzing for smart contracts is resource-intensive, making it hard to run comprehensive security testing at scale.

What I built

A fuzzing technique that uses information-flow guidance to focus fuzzing effort more efficiently, combined with static analysis to narrow the search space before fuzzing begins. Accepted for NGIRI funding and presented at the National Championship 2024, Islamabad.

  • Combined static analysis (Slither) with grey-box fuzzing (Echidna, Foundry)
  • Built a Streamlit interface for running and visualizing fuzz campaigns
  • Accepted for NGIRI funding and presented at the National Championship 2024, Islamabad

Researcher and developer — technique design, implementation, and evaluation.

PythonSoliditySlitherEchidnaFoundryStreamlit

Research complete, funded

Aurat March on Twitter: A Preliminary Look into the Disturbing Surge of Hate Speech

A research project analyzing hate speech, misinformation, and coordinated harassment targeting Aurat March discourse on Twitter.

Problem

Coordinated hate speech and harassment campaigns on social media are hard to quantify and characterize without large-scale, LLM-assisted analysis.

What I built

An analysis pipeline combining LLM-based data annotation, transformer-based classification, topic modeling, and network mapping to characterize the surge in hate speech and disclosure patterns around Aurat March discourse. Presented at DURS'24, Habib University, Karachi.

  • LLM-based data annotation using GPT-3.5 and GPT-4
  • Classification and disclosure analysis using BERT/RoBERTa and MiniLM
  • Topic modeling with LDA and conversation-network mapping in Gephi
  • Toxicity scoring via the Jigsaw API

Researcher — data annotation, modeling, and analysis.

PythonRBERT/RoBERTaGPT-4LDAGephiJigsaw API

Research complete, presented at DURS'24

SERVICES

What I can build

SaaS Platforms

Multi-tenant architecture, dashboards, financial/inventory tracking, secure data isolation between customers.

AI-Integrated Products

LLM-powered features (chatbots, classification, recommendations), RAG pipelines, workflow automation.

Full-Stack Web Apps

End-to-end builds from database to deployment, React/Next.js frontends, Python/Flask/FastAPI backends.

Automation & Integrations

CRM systems, SMS/communication workflows (Twilio), backend automation (N8N), third-party API integration.

Mobile Apps

Cross-platform apps with Flutter, from MVP to app-store-ready builds, including real-time features like maps and location tracking.

SKILLS & STACK

What I work with

Full-Stack Development

  • Python
  • JavaScript/TypeScript
  • Flask
  • FastAPI
  • SQL
  • React/Next.js

AI/ML

  • LangChain
  • RAG pipelines
  • OpenAI/Gemini APIs
  • Prompt engineering
  • BERT/transformer-based classification

Automation & Integration

  • N8N
  • Twilio
  • PostgreSQL
  • SQLAlchemy
  • Mailgun
  • APITemplate.io

Testing & DevOps

  • Selenium
  • Postman
  • Bitbucket CI/CD
  • Slither
  • Foundry
  • Echidna

Security-Minded Engineering

  • Multi-tenant isolation (JWT-scoped access)
  • Smart contract fuzzing/static analysis

CONTACT

Let's build something

Have a project in mind or want to talk through an idea? I usually reply within a day.

mmuaazulhaq@gmail.com+92 344 0687803
Punjab, Pakistan