Ramen journal · Lab · Çoruh Şimşek

SlurpSight, a web app for logging ramen bowls: an AI model reads the bowl photo, Google Maps finds nearby shops, and every review can be shared by link.

What it is

FIG. 1From bowl photo to shared review. The photo reading suggests the broth, the style and the toppings; the Maps search ties the review to a real shop and its location.
Source: SlurpSight code and design notes, Feb 2026

SlurpSight is a ramen journal on the web, live at slurpsight.web.app. You log a bowl with a photo, a rating and notes, and the review is tied to a real shop. Reviews can be public, each one has its own link, and a profile collects them. The design note for the last round put the use in one line: when a friend asks where to eat ramen in a city, the answer should be a link to your profile.

What happened

The first version came out of Firebase Studio in June 2025. I rebuilt it as SlurpSight in November and gave it a responsive redesign at the end of January.

On 15 February 2026 I audited the whole app and worked through the fixes in one day, 38 commits. The audit found real problems. The browser called the AI model directly, with the key in the page. The database had no rules file. And a new review saved the user's own GPS position as the restaurant's location. The AI calls moved behind three cloud functions, database rules and input checks went in, and a review now takes its coordinates from the Maps result or has none.

The same plan turned the app toward a personal journal: public profiles that work without signing in, shareable review links, dates for past visits and a data export.