EdTech / Tuition Marketplace·Nationwide marketplace
Building a location-aware matching engine for a nationwide tuition marketplace
How I built Tutorliy's radius-based tutor-matching algorithm for connecting tuition posts with nearby tutors across Bangladesh.
- Industry
- EdTech / Tuition Marketplace
- Client
- Tutorliy
- Timeline
- Nationwide marketplace
- Project
- Visit project
What changed
Before
After
Before
City/area name filters
After
Radius-based geolocation matching
Before
City-hardcoded logic
After
Nationwide matching architecture
Before
Naive full scans
After
Scalable geospatial queries
The challenge
Tutorliy needed to solve a matching problem that generic marketplace logic couldn't handle well: connecting students posting tuition requests with the right nearby tutors, at national scale, without manual moderation slowing every match down. A simple city-level filter wasn't precise enough — Dhaka alone spans neighborhoods where "same city" doesn't mean a reasonable commute. At the platform's volume of tuition posts and registered tutors, a naive matching approach would either miss good matches or slow to a crawl. The platform also needed to be trustworthy enough that parents felt safe sharing their location and contact details.
My approach
I built a geolocation-based matching system designed for scale:
- Radius-based geolocation matching — a custom algorithm that filters tutor-student matches by actual distance rather than city/area name, so a post only reaches tutors who can realistically take the job
- Nationwide architecture — built to work consistently whether the post comes from central Dhaka or a district town, not hardcoded to one city's geography
- Security-first data handling — student and tutor contact/location data protected with access controls appropriate for a platform handling minors' information and home addresses
- Scalable matching logic — designed so the matching algorithm stays responsive as tutor and post volume grows, rather than degrading into a slow full-table scan
The outcome
We shipped radius-based geolocation matching so tuition posts can reach nearby tutors nationwide, with query logic designed to stay responsive as the dataset grows.
Distance-based matching sounds simple until you have to make it fast at scale. A naive "calculate distance to every tutor for every post" approach falls apart as tutor and post volume grows — it means a large number of distance calculations per single match request. The core engineering work was designing the geospatial query logic so radius-based filtering stays efficient as Tutorliy's dataset grows, rather than slowing down as volume increases.
