Software Engineer

Muhammad Abdullah Amin

I like taking software from an idea to something running in the real world. Lately that has meant a lot of AI and retrieval systems, though what pulls me in is the problem, not the stack.

My WorkResume

About

Finishing things well enough to put them into the world.

Most of what I build starts with a problem I want to solve, or something I think could work better: a repetitive task, a slow workflow, a service that doesn't quite do what I need. Some of it stays small and focused. Some of it grows into a full system I design, deploy, and keep improving.

These days I work as an AI/ML engineer, so much of my recent work has been retrieval systems, LLM pipelines, and computer vision. I'd rather not be filed under one label though. I've written distributed shared memory in C, shipped Flutter apps, and built more command-line tools than I can reasonably justify. Give me an interesting problem and I'll happily pick up whatever it takes to solve it.

The satisfying part isn't figuring something out. It's finishing it well enough to put it into the world.

GitHub Activity

Abdullah's GitHub Contribution Chart

Work

Experience

Where I've been paid to solve the problem.

AI/ML Engineer @ Qult Technologies

Nov 2025 — Present

Lahore, Pakistan

  • An optimization system for plunger lift wells: time-series forecasting behind an autonomous controller that makes real-time calls to keep gas production up while limiting equipment damage and methane venting.
  • React and TypeScript dashboards for monitoring wells as they run, with a RAG chatbot layered on top so operators can ask a question in plain language instead of reading charts.
  • A computer vision pipeline for automating sand extraction, using thermal imaging and heat-pattern analysis with OpenCV and deep learning models, wired into robotic control.
  • Document intelligence and semantic search features built on LangChain, FastAPI, and vector databases.

AI/ML Intern @ Sysmatixx

Jun 2025 — Aug 2025

On-site

  • A smart inventory detection system for automated retail: dataset prep and augmentation in Roboflow, training in TensorFlow, deployed onto shelves that track stock in real time and trigger restocking.
  • A distributed lead generation platform, with a React frontend, Python services, Supabase for data, and n8n holding the automation together.

Stack

Core Technologies

What I reach for most often, though I'm always ready to pick up the right tool for the job.

Frontend

Frontend: React, Next.js, TypeScript, Tailwind CSS, Astro

Backend & Data

Backend & Data: Python, FastAPI, Node.js, PostgreSQL, Supabase, Docker

AI / ML

AI / ML: PyTorch, TensorFlow, scikit-learn, OpenCVLangChainQdrant

Systems

Systems: Rust, C, C++

Mobile

Mobile: Flutter, Dart, FirebaseReact NativeExpo

Selected

Projects

A few recent builds that show my range.

UniBot

A retrieval system for ITU that crawls the university's own sources and answers with citations, instead of guessing when the data has gone stale.

Python·FastAPI·RAG·Qdrant

Raah

A public transit companion for Lahore. Plan a trip across Orange Line, Metro Bus, Speedo, and E-Bus, then follow it step by step.

Flutter·Dart·Python·FastAPI

Winnow

An application intake and screening pipeline that scores candidates against a locked rubric and hands a ranked shortlist to a human.

TypeScript·React·Supabase·PostgreSQL

User-Level DSM

A user-space Distributed Shared Memory system providing transparent memory sharing across networked machines.

C·Systems Programming·POSIX·TCP/IP
View all projects

Contact

Get in touch.

Have a project in mind or want to collaborate? I'd love to hear from you.