06 — Web designer, full-stack developer & AI integration
tBoske — Snack Bar + AI Crowd Predictor
The problem: the owner often didn't know how busy it would get — meaning missed revenue and poor working conditions. The till data was already there, just unused. I extracted it, trained a predictive model on it, and surfaced it through a platform the owners use daily.
Role
Web designer, full-stack developer & AI integration
Tech stack
Next.js · TypeScript · Tailwind CSS · Python · Machine Learning
Year
2022 — 2026
Two projects for the same snack bar: first an informational website with menu and opening hours. Years later a platform that processes the shop's till data through a trained AI model — the model predicts how busy it will get based on historical patterns. The owner can now anticipate when extra staff or supplies are needed.
This project taught me that the most valuable data is often already present — the challenge is recognising and making it usable. Training an AI model on real business data is a very different exercise from an academic dataset: data is noisy, inconsistent and requires domain knowledge to interpret.