Case 01 of 03

VELORA

Fleet routing that doesn’t make you wait.

Kriti ’26 — bronze!

Dispatchtoday’s routessolving live

VehicleRouteStopsStatusAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA

16 stops in, 4 routes out — while the dashboard stays responsive.

Role

Full-stack developer

When

2026

Built with

Next.js · FastAPI · Celery · Redis · VROOM · OSRM · PostgreSQL

Result

Bronze at Kriti

The problem

“Delivery fleets juggle dozens of stops, vehicle capacities and time windows. Planning by hand is slow — and most solvers freeze the screen while they think.”

useless for a dispatcher watching the clock

How it worksSolving moved offthe request path.1Next.js UIZustand + Google Maps2FastAPIAccepts the job3RedisQueue + state4Celery workersRun jobs concurrently5VROOM + OSRMSolve on real roads6PostgreSQLPersist the routes↺ the client polls job status with TanStack Query

The hard parts

Stay responsive

Keep the API snappy while long optimisation runs grind in the background.

Real constraints

Model capacities and time windows without slowing the solve to a crawl.

Live status

Stream progress to the dashboard without hammering the server.

What came out of it

Bronze

at Kriti ’26, IIT Guwahati’s inter-hostel tech competition

Async

solve jobs run concurrently — the UI stays responsive under load

tools wired together, from Next.js to PostgreSQL

Life of a request

Velora · job ticket01

Queued

FastAPI

Takes the job, hands back a ticket in milliseconds.

IN LINE

Velora · job ticket02

Solving

Celery + VROOM

A worker picks it up and crunches routes on OSRM roads.

WORKING

Velora · job ticket03

Solved

PostgreSQL → UI

Routes saved; the dashboard’s polling finds them and draws.

DONE ✓

every request becomes a ticket — the UI just watches it move

View the code Read the notes

Next upWANDERLUSTAirbnb-style travel listings — Node, Express, MongoDB.

Back to work