Context
The ed-tech company's mobile app uses Sentry for crash reports, and by October 2026 the backlog was large. I was tech lead for an audit of it across 3 repositories.
The problem
There were about 1,000 open crash issues across the app, a video player fork and the native DRM SDK. Nobody knew which ones mattered, so none got fixed.
What I did
I triaged 1,000 issues across the 3 repositories, running the three tracks in parallel.
In the app, I fixed 14 groups: null guards, leaked animations, and a list crash on fast scroll.
In the player fork, I fixed Android foreground-service and concurrency crashes and iOS main-thread and teardown crashes.
In the native SDK, I fixed a download service that was not stopped on timeout and a logging format bug, and shipped a release.
I kept a ledger of the work, so what was fixed can be checked.
Key decision
I turned 1,000 reports into a small number of fixable crash groups across three codebases, instead of closing issues one by one. The crashes sat in different repositories, so I ran the tracks in parallel.
Result
1,000 issues triaged across 3 repositories, and 20 or more crash groups fixed.
What I learned
The audit does not yet show crash-free users before and after. Next, I would take that number at the start, so the effect is visible.
Stack
- Sentry
- React Native
- Kotlin
- Objective-C
- C
