Search like a shipper
Pick loading city, destination, vehicle type and length; results update as you go.
All work
Full stack · Next.js + MongoDB · Logistics
transport directory
Six printed broker books turned into a searchable directory of 2,276 Gujarat transporters, in days.
Overview
The brief
In Gujarat's transport hubs, the list of who runs which trucks to where lives in printed broker books passed from office to office. A shipper who needs a 32 ft container from Vapi to Delhi NCR flips pages and phones around. We turned those books into a directory you can search from a phone in seconds, with no commission and no middleman.
Problems & fixes
What was in the way
Pick a problem to see what was wrong, what we built and what changed.
The problem
Six city books (Ahmedabad, Ankleshwar, Baroda, Gandhidham, Surat, Vapi), photographed sideways, some pages tilted or blurred, with names, phones and routes in different column layouts per city.
What we built
An OCR pipeline on Apple's Vision framework plus a parser per city layout that pairs names, phones and routes, writes every uncertain row to a review file and checks it against the photo.
What changed
2,276 transporters from six hubs loaded and checked, covering 201 destinations and 10 vehicle types.
1 / 4
Impact
What it changed
2,276Transporters listed
6Transport hubs covered
201Destinations searchable
150 msWarm page load
- A shipper finds a matching truck for a route in seconds instead of flipping through a book.
- Transporters get enquiries directly, with no broker commission in between.
- The directory keeps itself up to date: owners claim and fix their own listings.
How it works
Under the hood
Click any part of the system, or play the flow from start to finish.
Step 1 · Photos
Printed books
Phone photos of each city's broker book, rotated and ordered page by page.
The result
Scroll through the site
Keep scrolling: the page inside the browser moves with you.
Claim your listing
A signed link opens a short wizard: check details, verify email, set a password.
Built in days
From the first photographed page to six cities live in under a week.
Timeline
- Day 1First book typed in, data model and search live behind a login
- Day 2Five more cities extracted with OCR; email sign-up and faster search
- Day 5Claim links, admin outreach queue and the public homepage
- OngoingCity-by-city outreach and listing updates
Tech stack
Next.js 16ReactTypeScriptMongoDB AtlasVercel (Mumbai)NodemailerApple Vision OCRPython
Tell us what you're building. A senior engineer, not a salesperson, replies within one business day.
Start a project