What it does
Employers post short videos instead of written job ads; candidates swipe through them. Onboarding asks what matters most to you, and that feeds a six-aspect matching engine which orders the entire feed by fit rather than by date. Like or dismiss a vacancy, save jobs, apply, and generate a formatted CV PDF from your profile. Swipy, an in-app AI assistant, answers questions about roles. Phone verification with country-code selection.
What I did
A video feed is the worst case for mobile performance, and this one was too slow to hold anyone. The fix was mostly about not loading things.
- Video players are initialised only when a card actually enters the viewport and torn down when it leaves, with the next video warmed in advance so a swipe never waits.
- Tracking so a user is never shown the same vacancy twice.
- Preference changes propagate through stream subscriptions and invalidate playlists in real time.
- Commercial ranking so paying clients surface first without breaking match relevance.
- CV PDF generation from the profile.
- The Swipy assistant, authenticated by posting a Firebase token into the webview.
- AppsFlyer attribution, and Crashlytics restructured so crashes became diagnosable rather than noise.
Result
80% performance improvement, 110% growth in daily active users, 3.7K monthly actives. The feed went from something people bounced off to something they came back to every day.
