Marrow
Engineering Manager, MobileIndia's leading app for NEET PG and medical exam prep.
Impact Log
What I have solved
Android lead for years, now Engineering Manager for Mobile across Android and iOS. Eight-plus years in Android, seven on this one flagship, an app past a million downloads. I pick up the surfaces nobody volunteers for, the parts where a bug quietly costs real money or breaks trust, and I own them until they are boring. Security work is described by outcome only.
Also supported the launch of DBMCI One, a NEET PG and FMGE prep app.
Play Store App Store
1M+ downloads
500K monthly actives
99.9% crash-free
300+ releases shipped
Surfaces I own
Secure video playback engine
core engine · 7 years- Own secure playback end to end: 2019 hackathon prototype to the engine the product runs on
- Offline download and play, multi-manifest streaming, subtitles, variable speed, resolution and edition switching
- Standardized error-code system for diagnosing field failures
Content protection and antipiracy
business critical- Led successive waves of DRM and antipiracy for premium medical video, the paid product
- The hard tradeoff: block leakage while paying users never feel the protection
- Outcome only, no methods shared
Mobile stability system
99.9% crash-free- Hold 99.9%+ crash-free for years across a huge, low-end-heavy device fleet on patchy networks
- Reproduce field crashes, ship targeted fixes: video threading, player memory leaks, lifecycle and binding races, OS-specific crashes
- Hardened cold start: successful launch rate 68% to 95%
Kotlin and Clean Architecture migration
~40% tech-debt cut- Migrated an inherited Java, MVP-era codebase to Kotlin, MVVM, and Clean Architecture, screen by screen, no big-bang rewrite
- Set the naming conventions and module boundaries the team still builds on
- Made the app genuinely testable
Payments and subscription flows
revenue-critical- Own the monetization surface: gateways, coupons, renewals for specialized exam tracks
- Handle purchase edge cases and eligibility checks where a bug directly costs revenue
How I ship Every rollout is a bet I verify. BigQuery event trends make the call, halt, scale, or roll back, and each ship gets measured after it lands, not just before.