In Dubai, I led the 0-to-1 build of OMS and WMS systems, moving from manual chaos to AI-driven predictability. Below is the data-backed narrative of that transformation.
30%
Productivity Lift
70%
Task Automation
40%
Error Reduction
Walking through the 50,000 sq ft facility, I saw pickers who looked exhausted but weren't actually getting many orders out. When I audited the floor data, the problem became glaringly obvious: they were spending 40% of their day just walking between aisles. Our fulfillment speed (Units-per-Hour) was flatlining even though we were hiring more people. We were throwing bodies at a problem that was actually a "geometry" problem—our software was asking people to traverse the entire warehouse for a single item, creating a massive bottleneck during peak hours.
I knew we couldn't just tell people to "run faster." We had to rewrite the logic of how we picked orders. I worked with the engineering team to move from "first-come-first-served" picking to Zone and Batch Picking. We built a system that grouped orders based on where the items actually lived in the warehouse. I also spearheaded the development of a custom Mobile WMS App to give pickers real-time, optimized routes on their handhelds. The goal was to make it so a picker could stay in one "zone" and maximize every movement they made.
The atmosphere in the warehouse changed. Productivity jumped by 30%, but the more meaningful metric was a 20% drop in worker fatigue reports. People were moving less but achieving more. For the business, our "Fulfillment Cycle Time" dropped so much that we started hitting a 98.5% SLA success rate. We turned a chaotic, high-stress floor into a streamlined operation where the software did the heavy lifting of planning, so the people could just focus on execution.
Units Per Hour (UPH) Baseline vs Post-Strategy
We were essentially "flying blind" with our inventory. During marketing promos, we’d sell out of popular items in hours (a 25% stockout rate), while slow-moving items sat on shelves for months, eating up capital. Our procurement team was using manual spreadsheets and "gut feelings" to decide what to order. It was a cycle of feast or famine: we either didn't have what customers wanted, or we had too much of what they didn't, leading to massive inventory carrying costs.
I launched an AI-driven forecasting model to replace the manual guesswork. We didn't just look at past sales; we built a multi-factor system that ingested market trends, upcoming promotional calendars, and even seasonal Dubai shopping patterns. I insisted on integrating these "Reorder Alerts" directly into the procurement workflow. The system wouldn't just tell us we were "low on stock"; it would predict exactly how much we needed to last the next 14 days based on predicted demand velocity.
Prediction accuracy shot up by 30%. The "human" win was seeing the virtual disappearance of stockouts during major Dubai sales events. No more frantic, last-minute emergency shipments at 3x the cost. The procurement team went from being "firefighters" constantly reacting to shortages, to strategic planners who could trust the data. It protected our revenue and ensured customers actually got what they ordered on the first try.
Accuracy Gain
+30%
Stockout Reduction
-50%
Our dispatch team was stuck in a data-entry nightmare. I watched as staff manually typed customer addresses and order details into carrier portals for nearly 70% of our orders. It was slow, mind-numbing work that led to a staggering 40% error rate—one typo meant a package went to the wrong side of the city. We were wasting thousands of man-hours on a task that felt like it belonged in the 1990s, and it was choking our ability to scale.
I decided to "kill the manual entry" by moving us to a headless OMS architecture. I led the technical integration of direct Carrier APIs. The goal was simple: the moment a picker scanned a package, a shipping label should pop out automatically with all the correct routing data already synced. I also co-led the negotiations with our shipping partners, using our move to API-driven volume as leverage to get better unit pricing and priority pickup slots for our drivers.
We successfully automated 70% of the manual shipping tasks. The dispatch floor went from a room full of people typing at keyboards to a high-speed transit hub. Dispatch errors dropped by 40%, meaning fewer angry customer calls and fewer "returned to sender" packages. Our delivery partners were happier because they received clean, real-time data, and our team finally had the headspace to focus on optimizing the last-mile experience rather than just avoiding typos.
MANUAL TASKS
-70%
ERROR RATE
-40%