# Secure Logistics ERP

> PAN India logistics platform

A logistics ERP built from scratch, handling 70,000 dockets a month for secure jewellery transport.

- **Page:** https://siddharthdeshpande.com/projects/logistics-erp
- **Built at:** MildlyClassic
- **Tech stack:** PHP, AWS, React, Node.js, MySQL

## Impact

- **2.5K** — daily active users
- **70K** — dockets/month
- **100K** — daily hits
- **0** — downtime during scale

## Problem

- Everything managed on spreadsheets and calls
- Manual coordination across multiple hub locations
- Phone-based delivery tracking only available
- Nothing connected end to end across operations
- Invoice reconciliation done manually in Excel
- No visibility into real-time shipment status

## Solution

- Full ERP system built entirely from scratch
- Finance, Ops, and Booking modules integrated
- Bulk docket creation via CSV file upload
- AWS migration completed in under 3 months
- Real-time tracking dashboard for all shipments
- Automated invoicing and payment reconciliation

## Outcome

- 2,500 daily active users across all hubs
- 70K dockets processed monthly without issues
- Zero downtime maintained during rapid scaling
- 100K daily API hits sustained in production
- End-to-end shipment visibility for all clients
- Manual processes reduced by over 80%

## How it works

1. **Operations** — Hub management, route planning, and real-time tracking across PAN India. Bulk docket creation via CSV upload handles 70K deliveries per month for secure jewellery logistics.
2. **Finance** — Integrated finance module handles invoicing, payment tracking, and reconciliation — replacing the parallel spreadsheets that ran the business before.
3. **Scale** — Grew from hundreds to 2,500 daily users with zero downtime. DigitalOcean to AWS migration completed in 3 months. Automated testing pipeline supports monthly releases.

## Key decisions

- **Monolith first, not microservices** — With a team of 7 and a product that needed to ship fast, a well-structured monolith was the right call. Finance, Ops, and Booking shared data models and transactions.
- **Full AWS migration over incremental hardening** — After a security incident on DigitalOcean, the call was to migrate entirely to AWS rather than patch. Three months of migration bought VPC isolation, managed databases, and a proper security posture.
- **Monthly release cycles from day one** — Established a 4-week build-test-deploy cadence early. This kept the team disciplined and clients expecting regular improvements rather than ad-hoc patches.

## What I'd change

- **Automated testing from month two** — The test suite came relatively late. For a system handling financial transactions and delivery tracking, automated regression tests should have been in place much earlier.
- **Cleaner multi-tenant architecture** — Adding new jewellery manufacturers as clients required more customization than it should have. A cleaner multi-tenant design from the start would have made onboarding faster.