Trailstory - One app to track your stories

🚀 Live at: Trailstory

Full-stack adventure tracker deployed on GCP (API), Vercel (frontend), and Supabase (PostGIS database).

Demo

Try the live demo at: Trailstory No signup required

Combine your multi-day adventure into a shareable page. Revisit your personal notes, photos, gear loadout and routes in one place. Get a day by day breakdown of your trip at the end.

Key Features

• PostGIS geometry types for route data & spatial queries
• Custom route aggregation logic for multi-day trip processing
• Custom authentication system (JWT tokens, rotating tokens, one-time tokens)
• Relational schema for domain entities: Users, Trips, Rides, Photos

Future Features

• Nearby trips: "Find trips within X km of this location"
• Photos near route: "Show photos taken along this route"
• Bounding box search: "Show trips in this map viewport"
• Strava / Garmin Webhook for trip syncing

Architecture

Production Architecture

• Frontend: React + Shadcn UI
• API: Python/FastAPI
• Database: PostgreSQL + PostGIS
• Storage: AWS S3 for image storage
• Deployment: Google Cloud Run

Design Decisions

Storing and Querying Geospacial data with PostGIS

GPS routes need efficient storage and querying for aggregation and map rendering. Rather than storing GPX files as text or JSON arrays, I converted them to PostGIS LINESTRING geometry types.

The trade offs were added database complexity and requires PostGIS extension, but enables:

• Spatial queries (bounding box calculation, point-in-route checks)
• Native geometry operations (route merging, simplification)
• Future features (nearby ride discovery, route similarity)

Trip aggregation

Users upload individual GPX files for each day's ride. Each ride's coordinates are parsed and stored as a PostGIS LINESTRING geometry in the database. Aggregation process:

• Query retrieves all rides for a trip, ordered chronologically by date
• Extract coordinates from each ride's LINESTRING geometry
• Concatenate routes end-to-start into a single continuous LineString
• Calculate aggregate statistics (total distance, elevation gain, highest point)
• Generate bounding box encompassing all rides

Why this approach:

• Preserves individual ride data for day-by-day breakdown
• Enables both per-ride and full trip views
• Aggregation computed on-demand, not stored redundantly

Authentication Strategy

I decided to implement custom JWT access tokens, Rotating refresh tokens and one-time token workflows rather than use libraries to better understand security and token lifecycles.