Shredder.

The original rider-matching app. Built and shipped solo.

My role: Sole engineer
Working with: Solo on the first build; a larger team by the time I returned
When: 2025 — initial build, then a later stint

Shredder / App Store artwork 1

Shredder / App Store artwork 2

The problem

Finding people to ride with is harder than it sounds. Ability is the thing that actually matters — a beginner and someone who only rides double blacks will both have a bad day — and none of that is visible in a group chat or a general-purpose social app. The brief was a way for riders to find other riders they would genuinely enjoy a day with.

What I built

My part
in the product.

Sole developer on the initial build, through to App Store release.

01
The whole matching system: a swipe-to-match flow between riders, and the pairing logic behind it.
02
Stat tracking for days on the mountain, so a rider profile is backed by what they actually ride rather than what they claim.
03
The full App Store release, and the bug-fixing pass that follows a first launch.
04
A web dashboard for the team behind the app.
05
Shredy, a matchmaker that scores riders against each other on stated interests and what they are looking for in a riding partner.

The current product has expanded beyond skiing and snowboarding. The artwork shows the current product; my initial solo delivery and later team engagement are distinct.

01 / a closer look

Find someone
to ride with.

I built the original swipe-to-match app and the pairing logic behind it. Ability and riding preferences were central to the original skiing and snowboarding brief.

02 / a closer look

From matching
to delivery.

The work included rider statistics, the App Store release, follow-up fixes and a web dashboard. My original solo build is separate from the later engagement with a larger team.

Decisions worth explaining

Why it works
this way.

Shredy is an algorithm, not a model. It is presented as an intelligent matchmaker and it is exactly that — deterministic scoring over stated interests and intent. Calling it AI would have been the easy marketing choice; it would also have been untrue, unpredictable for users, and far more expensive to run for a matching problem that does not need a model.

Ability first, not proximity first. Matching people who ride at the same level was the point of the product, so skill signals drive the pairing rather than simply who happens to be nearby.

Tools used on this project

React Native / TypeScript / Convex / Clerk

Next project