Delivery routes, staff rosters, production and pipeline schedules. I write your real operating rules into an optimization model (Google OR-Tools) and deliver a tool your team runs every day: a web app, an API or an Excel workflow.
Operators planned an industrial phosphate pipeline network by hand, 72 hours ahead: which feeders run, where the flow goes, when the pipeline pumps or stops, with every storage tank kept within safe limits and clients served around the clock.
I built a constraint-programming model of the whole network, about 10,000 decisions and 15,000 operating rules, with the pipeline's transit delay built in. Every schedule is checked by independent mass-balance validators and stress-tested by a simulator, and the planning systems request new schedules through an API.
| Before | After | |
|---|---|---|
| 72-hour schedule | Hours of spreadsheet work | About 15 seconds |
| Operating rule violations | Found after the fact | None, by construction |
| What-if (feeder outage) | Not possible | One API call |
Python, OR-Tools CP-SAT, FastAPI, Docker
Each runs in your browser with sample data, and each loads your own file.
Upload your stops, set your fleet and costs, and get the cheapest routes on real roads, with van capacity, delivery windows and shift length respected.
On the sample cities: 11–32% lower daily cost than a manual plan, often one van less.
OR-Tools vehicle routing, OpenStreetMap roads, Excel sheets and Google Maps links for drivers
Open the route optimizer
A weekly staff roster that respects rest times, contracts, leave, a senior on every shift and fair nights. Type a change like "Ahmed is off Thursday" and it re-plans in seconds, moving as few shifts as possible.
When a shift can't be covered, it says which rule blocks it instead of failing silently.
OR-Tools CP-SAT, plain-language changes read by an AI model, Excel export
Open the shift scheduler
Paste a website or upload PDFs: it answers only from that content, cites its sources, and says it doesn't know instead of guessing. Name the fields you need and it extracts them into a table.
Python scraping, search, AI answers, CSV and JSON export
Open the chatbotYou keep full control of the rules. I keep the model honest about them.
A spreadsheet, a whiteboard or "from experience" is a fine starting point. Send me a message on Upwork with the rules you plan by, and I'll tell you what an optimizer would change.
Arabic (native), English, French (basic). Based in Morocco, GMT+1, with good overlap with Europe, the Gulf and the US East Coast mornings.