APos

What it is

APos is the platform a pharmacy chain with 250 branches runs on. It is one monorepo of several apps that share one GraphQL API, built with React, TanStack Start and Electron, with AI features across them. I was the sole front-end developer, so I built every app and web front end in it.

The apps

• POS desktop app (Electron). The cashier's screen: scan a barcode, pick the LOT and unit, see the customer's details and health-insurance discount, take payment, open and close shifts, handle returns, and run stock counts. A second screen faces the customer.
• Admin. For the client, and close to a full business management system: it controls products, prices, branches, stock, staff, sales and reports for the whole company.
• Back office. For the APos team that runs and supports the platform.
• KYC system. Helps the client's HR team add employees.
• Scanning app. Reads LOT numbers from a photo with AI and fills in the stock.

Three problems I worked on

Pricing across +200 branches. Prices differ by city and region, so entering a price per branch meant +200 entries for every product. We created branch segments, groups of branches: a product's price is set once per segment and applies to every branch in it. This saved a lot of time.

Stock by LOT number. Staff typed every LOT number by hand when entering stock balances. I built the scanning app, where AI reads the LOT number from a photo and fills in the stock.

Stockouts. Managers did not always notice a product selling fast, so it ran out and the pharmacy lost sales. We built an AI recommendation system that uses last month's sales and seasonal conditions to warn the manager, for example "This product will sell out in 4 days".

Recommendations at the till

In the POS app the recommendation has two parts. The first looks at the customer's own purchase history, such as a medicine they buy regularly, and reminds the cashier they may need it again. The second looks at other customers: if people who buy product A often buy product B too, the app suggests B. This helps the cashier offer useful products and helps the pharmacy sell more. My part was the front end, how the recommendation is shown in the POS app, and work on its logic. I finished the feature, but it had not yet been rolled out to the branches when I left.