What I Built: A full-stack Flask application that tracks daily milk deliveries with automated billing calculations. Started as a simple tracker, evolved into a multi-tenant SaaS with user authentication, OAuth integration, and personalized settings.

Live: https://web-production-01a84.up.railway.app/

The Journey:

v1.0 - MVP (SQLite + Basic CRUD)

• Simple Flask app with local SQLite

• Deployed using Railway's GitHub integration

• Challenge: Database persistence across deployments

• Solution: Discovered Railway Volumes for persistent storage

v2.0 - Production Database Migration (SQLite → PostgreSQL)

• Hit SQLite's concurrency limits with multiple users

• Migrated to Railway's PostgreSQL in one click

• Challenge: Auto-increment sequences broke after data migration

• Solution: Created custom migration script to reset PostgreSQL sequences

v3.0 - Multi-User with Authentication

• Added OAuth (Google + GitHub) using Authlib

• Implemented email verification with custom token system

• Challenge: OAuth callbacks failing in production (http vs https)

• Solution: Dynamic redirect URI fixing:

v4.0 - User-Specific Data Isolation

• Added foreign keys linking records to users

• Challenge: Existing data had no user association

• Solution: Smart migration script that assigns orphaned records to primary user

v5.0 - Personalized Settings

• Per-user currency (10+ currencies) and milk pricing

• Auto-calculation with user-defined rates

• Optional bulk recalculation of historical records

• Challenge: Column additions without downtime

• Solution: Auto-migration on startup:

Railway Features I Love:

1. Zero-Config PostgreSQL: One-click database provisioning with automatic DATABASE_URL injection

2. Environment Variables: Seamless secret management across dev/prod

3. Volume Persistence: Used for backup storage before PostgreSQL migration

4. Automatic HTTPS: No SSL certificate headaches

5. GitHub Integration: Push to deploy - Railway handles the rest

Technical Highlights:

Smart Database Detection:

Automatic Schema Migrations: Instead of complex migration tools, implemented auto-detection and migration on startup - perfect for Railway's ephemeral deployments.

Session-Based Multi-Currency: Each user sees their own currency symbol throughout the app - achieved with user settings stored in PostgreSQL.

Architecture:

• Backend: Flask + SQLAlchemy ORM

• Database: Railway PostgreSQL with connection pooling

• Auth: OAuth 2.0 (Google/GitHub) + custom email verification

• Frontend: Jinja2 templates with responsive CSS

• Deployment: Gunicorn WSGI server on Railway

What I Learned:

1. Railway's DATABASE_URL uses postgres:// but SQLAlchemy expects postgresql:// - quick fix avoided hours of debugging

2. PostgreSQL sequences need manual reset after bulk imports

3. Railway's environment variable injection makes 12-factor apps trivial

4. Connection pooling is essential for Railway's PostgreSQL (pool_pre_ping, pool_recycle)

5. Auto-migrations on startup work better than separate migration scripts for small apps

Challenges Overcome:

• 404 errors post-deployment: Fixed with proper external=True in urlfor()

• OAuth state mismatch: Railway's HTTPS enforcement required callback URL fixes

• Data migration: Custom scripts handled SQLite boolean (0/1) to PostgreSQL boolean conversion

What's Next:

• CSV export functionality

• REST API for mobile apps

• Scheduled email reports using Railway Cron Jobs

Railway made deployment so smooth that I spent more time on features than DevOps. The PostgreSQL integration alone saved days of setup work. This project went from "weekend hack" to "production SaaS" thanks to Railway's simplicity!