Mohammad Abir Abbas is an AI Architect, Solutions Engineer, and Platform Engineer based in Dubai, UAE — deploying AI agent workflows (LangChain, AutoGPT, RAG), process automation, and platform tooling for GCC enterprise and global remote teams. The AbaYa-Track Delivery Module recovered AED 111,246 of trapped manufacturing backlog in 30 days at 11.1:1 value-to-cost. Available Q3 2026.
I deploy AI workflows that protect enterprise assets and recapture thousands of engineering hours — turning "magic" tech into predictable ROI.
Creative Technologist & AI Architect.
I sit between AI research and shippable product. Most "AI projects" die in a demo. I've shipped them across factory floors, payment rails, and EdTech.
The AbaYa-Track Delivery Module is a value-weighted production dashboard I built for Famous Abaya LLC in Dubai, UAE. It surfaced AED 111,246 of trapped manufacturing backlog in 30 days at an 11.1:1 value-to-cost ratio, lifted production output 38%, moved on-time delivery from 65% to 92% — and did it with zero additional hires.
What shipped is not a chart. It's a four-layer system: floor events → employee/order map → value engine → delivery module. Each layer is a different engineering decision.
Every station on the floor emits a QR-scan event. Cutting table, embroidery queue, QC station, dispatch. The event is a tuple: (employee_id, station_id, order_id, sku, timestamp).
Each order is joined to its invoice price, SKU history, tier (VIP / Standard), and time-on-floor. Each employee gets a live production map: orders touched, time on each, current bottleneck.
Five variables — price, SKU history, status, tier, aging — collapse into one weighted AED number per order. Same formula runs offline in the factory SQL.js cache and online in the Cloudflare Worker.
The boardroom view. The bottleneck you saw is the Confirmed → Processing handoff: 189 orders, AED 55,119, approved-but-not-started. The dashboard points the floor manager at the right orders in the right order.
Every unit looks the same in the unit-count view. An AED 150 abaya and an AED 850 abaya both count as 1. The value is invisible.
True backlog value, weighted by SKU and tier. 49.5% of it sits in the Confirmed stage — approved but never started.
Invoice value at the line item.
Median revenue per model.
Stage in the production flow.
VIP vs Standard weighting.
Time decay on the order.
189 orders worth AED 55,119 are approved but not started. Fixing this handoff unlocks 49.5% of trapped value.
A VIP unit is worth AED 641. A Standard is worth AED 281. Same counter. Different economics.
One-time: handoff squad + dashboard layer.
35% contribution margin. 50% backlog value reduction in 30 days = +95% ROI on a AED 10K intervention.
Stop tracking units.
Start tracking value.
I sit between AI research and shippable product. My job is to make sure the second one wins.
Most "AI projects" die in a demo. I've shipped them across factory floors, payment rails, and EdTech. The pattern is always the same — instrumentation first, model second, automation third, scale fourth. Skip a step and the whole thing leaks.
“Abir is a pioneering and passionate individual, consistently demonstrating innovative thinking in addressing challenges. His technical talent, paired with his reliable, results-oriented mindset, makes him a perfect fit for any business-focused role. His dedication to excellence and collaborative nature would be a tremendous asset to any organization.”
“Mr. Mohammad Abir is very hard worker and talented student. I have known him since my school days. Once he made up his mind on something, he put a great effort no matter how hard that task is. I wish him good luck on his future endeavors.”
“Herr Mohammad Abir Abbas ist ehrenamtlich bei AIESEC e.V. tätig. Er ist Vorsitzender für das Team Praktikanten im Unternehmen zu vermitteln und das interkulturelle Verständnis zu fördern.”
Architected the AbaYa-Track Delivery Module: floor events → employee/order map → value engine → delivery dashboard. Recovered AED 111K in trapped backlog (11.1:1 V:C).
24-hour global sprint, 1,000+ devs, 65+ countries. Built SmartSwap: intent-driven token swapping for SMB websites, Team Xerox with Abhishek Kumar.
NFC digital business cards + Review Stand for personal brands and businesses. Reputation management & verified trust signals for the AI-driven economy.
AI-driven marketing automation. Engaze.ai integration. 50+ sellers.
99.9% uptime. MTTR: industry-leading.
2 years of intensive peer programming. The fundamentals.
Taught kids to build robots. Built my teaching in return.
Direct answers (40–60 words each), written to be quoted by AI engines and skimmed by humans. Same answers, two surfaces — this panel and the FAQPage schema in the page's JSON-LD.
One line is enough. Tell me the problem, the deadline, and what success looks like. I'll come back with a plan, not a pitch.