• 

  landing Page

Bidangil – Bringing Korea’s Storefronts to the World

🚀 Inspiration

As a Korean-American living in the U.S., I constantly hit the same wall: Korean online stores insist on a local phone number + credit card and seldom ship abroad. Friends were my “human API,” placing orders on my behalf—an experience shared by thousands of immigrants and K-culture fans worldwide. Bidangil was born to remove that friction and keep people engaged through a built-in community hub while we bundle their orders into one low-cost shipment.

🛠️ What We Built

• Full-stack web service:

  • Frontend – Next.js 14 (React 18) + Tailwind; real-time UX via WebSockets
  • Backend – Django 5.2, DRF, Django Channels, Celery, Redis, MySQL

• Google Maps API integration: a static map snapshot appears inside a two-panel order form so shoppers instantly verify their delivery address as they type.

• Cloud infra: GCP for object storage & global CDN, plus GitHub Actions CI/CD.

🏗️ How We Built It

1. Order & payment automation – Webhooks drive a two-step Stripe flow (items → delivery) triggered by a custom save_formset override so exactly one payment record is created per order.
2. Real-time pipeline – Long-running avatar generation off-loaded to Celery; when the image is ready the backend pushes it to the browser over Channels, cutting web-worker hold time from 30 s to 0.05 s and raising sustainable concurrency from 8 → 200+.
3. Comment trees at scale – Optimized queries (select_related + prefetch_related) and an in-memory lookup table serialize nested comments in O(N) without extra DB hits.

📚 What We Learned

• Maps ≠ maps – Even a tiny Static Maps thumbnail meaningfully boosts address-entry confidence and reduces checkout errors.
• UX is compound interest – Micro-interactions (live nickname checks, sticky avatar quiz progress) translate directly into session length and conversion.
• Back-pressure matters – Moving IO-bound tasks to Celery plus channel-layer pushes kept the API responsive under spike traffic.

🧩 Challenges & Solutions

A quick math peek

We estimate delivery upfront with

$$ \text{Total Cost} = \left(\sum_{i=1}^{n} \text{KRW}i\right)\times r \;+\; w \cdot c{\text{kg}} $$

where $r$ is the live ₩→\$ rate and $w$ the consolidated weight. This transparent formula demystifies cross-border fees for first-time buyers.

🔭 What’s Next

• Dynamic Place API look-ups to auto-suggest Korean store locations.
• Regional fulfilment hubs to shrink $w$ in the equation above.
• Open-sourcing our Django Channels patterns so other small shops can go global.

Bidangil turns “Can you order this for me?” into a single click experience and Google Maps made the last-mile feel as tangible online as it is on the street.

Built With

• celery
• django
• fedex
• gcp
• google-maps
• next
• nginx
• pm2
• postgresql
• react
• redis
• stripe
• websockets