Jul 2024 — Dec 2024
CartBliss
A unified e-commerce and inventory management platform for a local supermarket, cutting manual task time by 40% and stockouts by 60%.
- Django REST Framework
- PostgreSQL
- Bootstrap
- DevOps
Problem & context
A local supermarket was running sales and inventory as separate, manual processes — stock counts and online listings drifted out of sync, which led to both wasted staff time and regular stockouts on popular items.
Built end-to-end for a real local supermarket: one platform covering both the customer-facing storefront and the inventory management the staff actually use day to day.
Constraints
- Had to serve up to 200 concurrent users without degrading response time
- Inventory changes needed to reflect in the storefront close to immediately
- Non-technical staff needed a workable, low-friction interface
Trade-offs
- Chose Django's MVT pattern and PostgreSQL for a batteries-included path to a consistent schema, over assembling a more custom stack that would have taken longer to harden.
- Prioritized query performance (schema design, indexing) early, since inventory and checkout flows are read-heavy and any lag there is felt immediately by staff and customers.
System flow
Django’s MVT pattern separates the storefront views, inventory models, and templates cleanly, with PostgreSQL as the single source of truth for both stock levels and order data — so a sale and a stock adjustment are always consistent with each other.
Screenshots
Add screenshots of the storefront and the inventory dashboard here.
Results
- Query execution times under 300 milliseconds
- 1.5-second response time maintained for up to 200 concurrent users
- Manual task completion time reduced by 40%
- Monthly stockouts reduced by 60%
Lessons learned
Getting the database schema right before writing views saved significant rework later — most of the performance headroom came from schema and query design, not from optimizing at the application layer after the fact.