<aside>

Software Engineer, Devsinc · Jun 2022 to Aug 2023

</aside>

About Leaf

Leaf was a private voice-memory app for iOS and Android that helped families preserve, organize and privately share recordings of loved ones.

I worked across the Flutter app's architecture and core recording and playback experience, including performance improvements, AI transcription, synchronized transcripts, offline playback and private sharing.

At a glance

<aside>

50% faster repeat-visit load times · Flutter · Whisper transcription · Word-level audio sync

</aside>

• Led 2 engineers while moving the app toward MVVM
• Introduced caching, lazy loading and centralized application state
• Built Whisper-based transcription with synchronized word-level playback
• Shipped offline playback and private sharing through deep links

The Problem

Leaf had grown beyond a simple audio recorder.

Users were recording personal memories, returning to old recordings, reading transcripts while listening, and sharing those memories privately with family. As the product grew, the mobile codebase and data flow became harder to maintain, while repeat visits to existing content were slower than they needed to be.

The challenge was to improve the app's internal structure and performance without stopping feature delivery on the recording and playback experience.

What I Did

I led a gradual move toward MVVM rather than trying to rewrite the application all at once.

The work separated UI, application state and data access more clearly so features could evolve without pushing more logic into screens and widgets.

Alongside the architecture work, I introduced: