Accelerating on-device ML on Meta’s apps with ExecuTorch
WhatsApp/Messenger moved key models on-device; reduced model load & inference time and improved ANR metrics. (Engineering at Meta)
Showing 20 of 32 articles (Page 1 of 2)
WhatsApp/Messenger moved key models on-device; reduced model load & inference time and improved ANR metrics. (Engineering at Meta)
Case study + note that ADPF is broadly available (Unity/Unreal/Cocos/native) to tune thermal/perf. Last updated Jul 9. (Android Developers)
How over-instrumentation hurts startup, memory & battery; pragmatic sampling/SDK tips. (droidcon)
Coinbase fine-tunes network calls to make their app feel snappier for users.
Gojek tells how they upgraded from basic bid alerts to the slick ‘Courier’ system.
DoorDash uses Flink to spot user sessions and send timely notifications.
Gojek’s ‘Courier’ service makes push notifications quicker and more reliable.
Instagram tunes DMs to be quick and trustworthy for every user.
Glance uses a CDN to build a quick read-through cache for speed.
Flipkart cuts mobile page load times with caching and lazy tricks.
Pepperfry uses Java multi-threading to make API calls lightning-fast.
Instagram fetches data early in the background to keep feeds flowing.
Swiggy’s search delivers results fast to keep users satisfied.
Zomato rewrote their Android network layer for faster, better flow.
Grab digs into high latency in their Market Store and fixes it fast.
Meetup scales their notification system to handle queues like a boss.
OkCredit slashed Android cold startup time by 70% with clever fixes.
LazyPay speeds up media delivery with CDNs for a snappy fintech app.
Tokopedia hashes GraphQL queries to shrink payloads and boost speed.
Dropbox sped up Android camera uploads and made them rock-solid.