WINE

One-liner: A social wine tracking & discovery app — snap a bottle or a glass, tag where you’re drinking on an interactive map, rate it, and see what your community is uncorking nearby, backed by a ~50,000-winery location database.

Origin (the Build Loop)

Because I wanted to remember what I drink and where, I kept hitting the fact that every wine app (Vivino, CellarTracker, Delectable) makes you scan a bottle label and writes a text note — none of them map where you drank, and none can identify a wine from a photo of the glass in your hand, so I built WINE: photo ID for bottles and glasses, a Leaflet map of every tasting, and a community feed, that turns “what am I drinking” into “where should I drink next.” It taught me that a domain “everyone has an app for” can still have a whitespace — the map + glass-ID + social combo simply didn’t exist, and free data sources (OSM Overpass, TTB, WhiskeyFYI) were enough to fill it.

The problem

The wine app market has 70M+ users across Vivino, CellarTracker, Wine-Searcher, and Delectable — yet all of them miss the same three things:

1. No location map — text notes at best; no interactive map of everywhere you’ve tasted or what people near you are drinking.
2. No glass photo ID — every scanner requires a bottle label, but the moment you actually want to log a wine is when it’s already poured.
3. No real social discovery — Vivino’s feed is a marketplace; Delectable (the best social app) was abandoned.

What it does

• Photo identification for bottles (label scan) and glasses (color/legs/opacity heuristics).
• Interactive map (Leaflet + OSM, marker clustering) with personal/all/winery/distillery modes and an interactive home mini-map.
• ~50,214 wineries across 60+ countries in the locations database, with ~33% carrying rich descriptions from venue enrichment (website crawl + Google Places).
• Distillery coverage — imported via TTB permit data + OSM Overpass + WhiskeyFYI.
• Spirit groups — create/join groups, group-filtered feed, category model.
• Ratings, reviews, venue detail pages with stats, wines poured, and recent tastings.
• Notifications and security hardening alongside the map improvements.
• Premium design system (Sept 2026) — brand identity, expanded color palette, depth and material treatment across every page.

How it’s built

• Stack: Python (FastAPI), SQLite (WAL mode), Leaflet + OpenStreetMap, vanilla JS frontend; ~608K lines across Python/HTML/CSS/JS.
• Notable engineering:
  • Free-data venue pipeline — OSM Overpass queries (via curl subprocess after httpx kept failing) + TTB permit data + WhiskeyFYI imports built the 50k-winery base at zero cost; website crawling + Google Places enrichment layered descriptions, phones, and images on top (+412 wineries, +505 crawled sites in one batch).
  • WAL-mode SQLite at scale — one file backing a map app with thousands of location rows and live check-ins.
  • Glass photo ID — the feature none of the 70M-user incumbents ship.
• Architecture: FastAPI backend → SQLite (locations, wines, tastings, users, groups) → Leaflet map + server-rendered pages. Runs on :8002, live and actively developed.

Proof points

• ~50,214 wineries, 60+ countries — a free-data venue database that would cost real money from a commercial POI vendor.
• 16,886 rich venue descriptions (33%) via the enrichment pipeline.
• 291 distilleries imported from a single fixed Overpass query batch.
• Live on :8002 and iterating daily — groups, notifications, security, and map modes all shipped in its first week.

What to show

• Demo: The running app (map with clustered winery pins, a venue page, a tasting check-in flow).
• Visuals needed: the map view with clusters; a glass-photo ID result; a venue detail page with enrichment; the group feed.

Cross-links

• The map + geocoding + POI aggregation DNA is shared with k9-overwatch (lost pets on a map) and civic-duty (civic data on a map).
• The free-data sourcing discipline (TTB/OSM/government sets) is the same muscle as api-access and infra-metals-dashboard.
• Photo-ID-of-objects is the pattern from thrift-lens and complete-paperwork pointed at wine.

Case-study angle

Alex looked at a 70M-user market, found the exact three features every incumbent was missing, and built them on free government and OSM data — a 50k-venue discovery platform in its first week of life. The whitespace thesis wasn’t a guess; it’s the product.