A-01About
Hi, I’m Alaa.
Software engineer who builds full systems end to end — front end, backend and data — and now focuses on multi-agent LLM systems that run in production.
Damascus, Syria5+ YRSUI → AIEN / AR


01
The story so far
From UI to AI.
- 2021
I started in 2021 as a junior freelance developer. I built web apps for small businesses end to end — requirements, database design, implementation and deployment — in PHP, Laravel, Vue.js and MySQL. It was my first time working directly with clients: working out what they actually needed, shipping to deadlines and supporting what I’d delivered.
- 2022
In 2022 I joined AL-MOTKAMEL Global Co. in Damascus as a core engineer on Minaret, a complete ERP platform rather than a single system. I worked on the shared core, not on isolated features, so every change had to hold across the whole platform. I worked the data layer at multi-million-row scale: composite indexes, partitioning, query rewrites, and heavy reporting moved into stored procedures. I also led the full redesign of Akar, a real-estate management system.
- 2024
In 2024 I finished my degree in Information Technology Engineering at Al-Jazeera Private University and started a Master’s in Web Science & AI at the Syrian Virtual University. That year, at Be In Media, I took over an HR and payroll module that was producing wrong results and brought it to a stable state. I did it inside an established Laravel HMVC codebase, following the conventions already in place.
- 2024
From May 2024 I freelanced for a year and delivered three production systems solo: a clinic management system, a transportation management platform and an enterprise CMS. I owned every stage, from requirements and schema design to deployment and post-launch support. My clients weren’t technical, so the job was turning how they actually worked into a system they could run themselves.
- 2025
Since May 2025 I’ve owned the backend and AI layer of Muqnea, a live generation platform: Python microservices, queue-backed workers and a multi-agent pipeline where specialised agents hand off schema-validated output. Generation cost is down 82%. Three changes did that together: routing each step to the smallest model that passes its eval, caching stable prompt prefixes and running steps in parallel. First output now arrives in 15–30 seconds, with quality measured at 91% on our internal rubric.
- 2026 · NOW
For a year alongside that, I also led the front end of Pazar Pro, a multi-vendor marketplace, through to handover in August 2026. Today my focus is multi-agent LLM systems that run in production, measured with real numbers rather than impressions. I write down what actually works in Production Notes, a series I publish on LinkedIn in English and Arabic.
02
The rules I work by
Short, testable, and learned in production. Each one comes with a number or a note I wrote about it.
01
Judge it by the 100th run, not the first
One great answer doesn’t make a prompt finished. It makes it lucky. I call a prompt or agent tuned when about 98% of results meet my spec over many runs with real inputs, on the same model. Then I try a smaller, cheaper model. If it still hits 85–92%, it’s robust.
02
Numbers beat adjectives
“Much cheaper” is an opinion. “82% lower generation cost” is a result, and it came from three changes together: routing each step to the smallest model that passes its eval, caching stable prompt prefixes and running steps in parallel. Quality and speed get numbers too: 91% on our internal rubric, first output in 15–30 seconds.
03
Work in the language you can verify
I prompt in the language I can check the answer in. The risk isn’t the prompt, it’s the reply: a plan full of half-understood terms that you accept anyway. I hold agents to the same rule. They hand off schema-validated structured output, not free text, so every step can be checked.
04
Fit the system, not my habits
In an existing codebase, I follow the conventions already in place instead of adding my own. On Minaret, I worked on the shared core of an ERP platform, so every change had to hold across the whole platform, not just the screen in front of me.
03
How I work
Back end, data and front end, plus the multi-agent AI layer that now sits on top of them.
- 01
Start with how people actually work
I talk to the people who’ll run the system, technical or not, and scope what’s actually needed before I write code.
- 02
Model the data
I turn the business process into a relational model and clear workflows, planned for real volume: indexes, partitions and caching where they’re needed.
- 03
Agree the API contract
Endpoint shape and response contracts come first, so the front end and backend build against the same thing. APIs are versioned, so they can change without breaking the apps that use them.
- 04
Build the interface people run
Screens scoped to each role, kept fast as the data grows: pagination, virtualised lists and client-side caching.
- 05
Measure it
Every AI step has an eval, quality gets a rubric, and cost and speed get real numbers. One good run isn’t proof.
- 06
Ship it and stay
I deploy, hand over and support what I deliver after launch.
04
Education
2024 – Present
Master’s Degree · Web Science & Artificial Intelligence
Syrian Virtual UniversityIn progress
2019 – 2024
Bachelor’s Degree · Information Technology Engineering
Al-Jazeera Private University
05
Quick facts
- Based in
- Damascus, Syria
- Experience
- 5+ years in production systems
- Currently
- Backend & AI Systems Engineer · Muqnea
- Focus
- Multi-agent LLM systems in production
- Writes in
- English and Arabic · Production Notes
- Education
- Master’s in Web Science & AI (in progress)
Here’s my CV and every way to reach me. Email is the fastest.