Abdullah Saleem
Full-Stack Developer & Computer Vision Engineer
Full-Stack Developer & Computer Vision Engineer
I build the software a business actually runs on.
Admin platforms and ERP, B2B sites and storefronts, and the automation that removes the manual work between them. And when a product needs to see, I build the computer vision inside it: real-time detection that ships on real hardware, not a notebook.
- Yrs full-stack
- 4+Yrs full-stack
- Yrs computer vision
- 3+Yrs computer vision
- Projects shipped
- 120+Projects shipped
- Design to deploy
- End to endDesign to deploy
Shipping software since 2022. Find me on:
Software that went into production and stayed there.
8 projects. Swipe to browse.
Full-Stack Systems2026Fenix Brokers
A full B2B platform for a cosmetics broker: product catalog, quote requests, an admin panel and email campaigns, all in one place.
+50%client inquiries in 2 monthsNext.js · React · TypeScript · Supabase
Case study
Computer Vision & Platform2026PPE Monitoring Vision System
Site cameras that check every worker is wearing their hard hat and vest, and a dashboard where a safety team watches compliance across every site and gets alerted the moment someone is not.
Multi-sitelive compliance monitoringYOLO · PyTorch · OpenCV · Next.js
Case study
Multimodal / VLM2024Virtual Try-On
See how a garment actually looks on you before buying, rendered by a vision-language model that reasons about fit.
VLMvision-language try-onPython · PyTorch · VLM · Diffusion
Case study
Full-Stack Systems2025Animated Film Network
A daily animation news site, rebuilt with a fast modern front end over the publisher's existing backend.
Dailypublishing, no rebuild stepVite · React · PHP · REST
Case study
Computer Vision2025MagicQC
A desktop app that measures garments from a camera to 0.2 cm, so quality control stops being slow, manual guesswork.
0.2 cmmeasurement accuracyOpenCV · Python · Laravel · MySQL
Case study
Automation & CRM2025Patient Lead Routing
Patient enquiries reaching a hospital group on WhatsApp, forms and calls, answered in Spanish by a bot and booked onto the calendar with no manual typing.
Zeromanual data entryn8n · WhatsApp API · GoHighLevel · LLM routing
Case study
Computer Vision2024UAV Search & Rescue
A search-and-rescue drone that spots people from the air on its own, deciding on-board in real time instead of streaming to a server.
Real-timedetection on the edgeROS 2 · YOLOv8 · ONNX · PyTorch
Case study
E-commerce2024Palazzo del Benessere
An online store for a wellness practice, selling products, courses and remote consultations to an international audience.
3 in 1products, courses, consultationsWordPress · WooCommerce · PHP
Case study
11 more in the archive: e-commerce, business sites and internal platforms.
What the people who paid for it said.
Client feedback on work I delivered, solo and with a team.
Three lanes, one pipeline.
Most projects fail at the seams: between the product and the model, or between the software and the box it has to run on. I own the whole run, which is why nothing gets handed over half-finished.
Full-Stack & Product
The software a team runs the business on: admin platforms and ERP, B2B catalogs and quote flows, storefronts, dashboards and internal tools. Built to be handed over and extended, not demoed once.
- Next.js
- React
- TypeScript
- Supabase / PostgreSQL
- Laravel
- Tailwind
Computer Vision & Edge AI
Detection and tracking that has to hit frame-rate on constrained hardware. Training and data through to quantized models running on the device itself, not a round-trip to a server.
- YOLO / YOLOv8
- PyTorch
- OpenCV
- ONNX
- TensorRT
- Jetson / Raspberry Pi
Automation & Deployment
The plumbing that removes manual work and keeps things running unattended: inbound-to-CRM pipelines, LLM routing, containers, and the infrastructure underneath.
- n8n
- LLM routing
- Docker
- ROS 2
- nginx / Redis
- AWS / Vercel
I ship products, not demos.
I am Abdullah Saleem, a full-stack developer with about four years of shipped work behind me. Most of what I build is the ordinary, load-bearing kind: admin platforms and ERP, B2B catalogs and quote flows, e-commerce, internal tools, and the automation that takes the retyping out of a team’s week. Next.js, React and TypeScript on the front, Supabase, Postgres, Laravel and n8n behind it.
For about three of those years the work has also been computer vision, and that is the part most developers do not do. Object detection and tracking that has to hold frame-rate on constrained hardware (YOLO, PyTorch, OpenCV, ONNX, Jetson and Raspberry Pi), from data and training through to Dockerized deployment.
On MagicQC I put an end-to-end industrial CV system into production: garment measurement to 0.2 cm from a camera, with the desktop app operators use and the cloud platform managers report from. On a PPE monitoring system I built both halves again: per-worker hard-hat detection, and the multi-site dashboard a safety team runs it from. That pairing is the pattern. The model is rarely the deliverable, the product around it is.
What makes me useful: I can take something from a blank repository to a running product a team actually uses, and I am not stuck when the hard part turns out to be a camera.
Need software built, or a system that has to see?
Tell me what you are trying to build, automate or detect. I will tell you honestly whether it is worth doing, and how I would take it from nothing to something your team runs every day.
Open to full-stack and computer vision engineering roles (remote or Islamabad), and to select freelance work. I usually reply within a day.
Four years building full-stack software, three of them on computer vision, with systems running in production today.