GameShelf

Every game group ends up asking the same questions: which games do we own, who wins most often, what did we play last time? GameShelf answers them with a full multi-user product: collections, photo-logged sessions, stats, and anonymised community charts.

The crowd-pleaser: photograph your game shelf and Gemini Vision detects the titles, adding them to your collection after you confirm.

What it does

• Personal game collection and a session log: players, winners, duration, photos
• AI shelf scanner: one photo → detected titles → confirm → collection populated
• Personal statistics plus community top charts from anonymised aggregates
• Full trilingual UI (English, Romanian, French)
• CSV export and GDPR-style account deletion

Engineering highlights

• File storage is a pluggable abstraction (local disk / Azure Blob) — cloud migration is an env-var flip, and photos stream through the API to preserve per-user authorization
• Azure infrastructure lives in 5 reusable Terraform modules, with a scale-to-zero staging environment powered on/off on a schedule
• Async end-to-end backend: FastAPI + SQLAlchemy 2.0 async + asyncpg
• GitHub Actions CI/CD with OIDC federation to Azure — zero static credentials