I'm Hassan Ali, a full-stack and AI engineer. Devrxy is the studio I ship my own work under, and "end to end" is meant literally: the engine, the AI, the servers it runs on, the interface, and the store listing at the end of it.
One app out in the world today, with more in the pipeline. Everything here was built end to end rather than assembled.
A tactical duel for two. Race your pawn to the far side while walling your opponent in, but never quite sealing them off. No dice, nothing hidden, most games over in a couple of minutes.
The studio ships one thing at a time and finishes it. This slot fills in when the next one does.
Full-stack engineer. Flutter and AI/LLM products, RAG and vector search. Lahore, working remote worldwide.
Five years, fifteen shipped apps, and a stubborn interest in the last ten percent. Most of my work is Flutter on the front and TypeScript, Node or Python behind it. These days a lot of it is AI: retrieval pipelines, embeddings, and getting model costs down to something a business can actually run.
I lead a team of five at Brainium Technologies, where we ship weekly across three SaaS and AI products. I built AskLumo's retrieval path end to end, from ingestion and chunking through pgvector search and FastAPI, and took OpenAI, Claude and Gemini into production behind it. On VoxBox I got a session down from about $11 to $1.60 by cutting repeat calls and caching what didn't need asking twice.
Before that I was writing Flutter at Boolean Bites, where Imagitor passed a million downloads and Houzi turned into a white-label platform for a 56,000-strong real-estate ecosystem. On Marham, Pakistan's largest healthcare platform, we took page load from four or five seconds to 1.2, put conversions up 30%, and halved the infrastructure bill.
Devrxy is the other half: my own apps, built without a client brief. Gridlock is the first one out.
Built for employers and clients, not under Devrxy. Here because the numbers are the point.
Open to remote roles worldwide. Comfortable across EMEA, US and APAC hours.