Optimizing PPE dispensers

Many people saw a stock problem. I saw a space problem. Every slot in a dispenser represents a cost — the machine itself, the space it takes up, its limited storage capacity. A low-consumption reference sitting in a slot is a slot that can't serve a reference in higher demand. Looking across several dispensers, I noticed some were always full while others kept running out. The issue wasn't the stockroom clerk, or even consumption levels: it was existing capacity poorly distributed, with no simple way to check it.

I built a recommendation engine that automatically compares, for each reference, its actual consumption against the number of slots currently assigned to it. The algorithm never tries to add slots: it redistributes the ones that already exist. Within a given slot type, each reference is assigned a share of total capacity proportional to its real weight in consumption — freeing up slots on low-demand references and reallocating them to fast-moving ones, without ever changing the dispenser's physical capacity. The goal isn't to add dispensers or more slots: it's to make every existing slot pay for itself through an optimized layout based on real consumption. This application doesn't replace field expertise: it provides objective recommendations to help design dispensers that perform better and are easier to maintain.