Graduation project · Jan – Aug 2026
IntelliPharma
Planning, selling and delivering medicine from one platform
Backend engineer: owned route planning and the AI-optimizer integration, co-built the ERP core, designed and ran the evaluation
- median shorter routes vs. list order
- 43.9%
- median gap to the exact optimum
- 8.1%
- stop coverage on 79-stop days after the fix
- 24% → 100%
- Problem
- pharma distributors run their warehouse, sales reps and delivery drivers on separate tools, spreadsheets and memory, so routes are guessed, visits are self-reported, and stock, cash and targets get reconciled at the end of the month, a problem validated with two pharmaceutical warehouses and field pharmacists in Syria.
- Solution
- one modular Laravel backend that holds the catalogue and stock, books and prices orders, plans every rep's and driver's day with an AI optimizer, replans mid-day, and sends what really happened back to the optimizer every night, with driver identities pseudonymized.
- Gains
- in a 117-run evaluation against the production API, optimized driving tours were a median 43.9% shorter than visiting stops in list order and came within 8.1% of the mathematical optimum. The evaluation also surfaced a bug, and after the fix, coverage on large days went from 24% to 100%.