TechPrufer
Get started
Grab
Grab logoGrab · Feature trail

Android App Size Optimization

4 posts · Feb 2020 – Mar 2026

What this trail covers

Grab has implemented Project Bonsai to systematically reduce Android app download and disk sizes. This initiative focuses on measurement, reduction, and containment strategies. Key technical efforts include developing a custom analytics tool (App Sizer) for detailed app binary analysis, optimizing DEX file size by addressing transitive R classes and third-party library bloat, and refining resource management through techniques like using vector images and WebP format, and enabling language resource optimization. This post details the successful enablement of R8 optimization for the Grab Android app, achieving significant improvements in app size, startup time, and stability through innovative AI-assisted debugging techniques, a pragmatic testing strategy, and an optimized feedback loop.

All Grab trails
  1. Enabling R8 optimization at scale with AI-assisted debuggingThis post details the successful enablement of R8 optimization for the Grab Android app, achieving significant improvements in app size, startup time, and stability. The core challenges addressed were slow debugging of obfuscated/optimized code, the difficulty of comprehensive testing at scale, and a slow feedback loop due to long build times. The solution involved three key innovations: 1. AI-assisted debugging using custom MCP tools for APK decompilation and stack trace deobfuscation, integrated with GitLab CI to create multiple parallel merge requests for testing different solutions. 2. A pragmatic testing strategy that involved a phased rollout and a QA build configuration matching production. 3. An optimized feedback loop by leveraging remote CI builds and AI to manage parallel testing and verification. The post highlights the impact of AI in accelerating investigation and verification cycles, reducing manual reverse engineering, and enabling parallel experimentation.
  2. Android App Size at Scale with Project BonsaiThis post details the 'Project Bonsai' initiative to reduce Android app size. It outlines the strategy of Measurement, Reduction, and Containment. Key technical contributions include the development of a custom analytics tool 'App Sizer' for detailed app binary analysis, which integrates with Grab's CI system and sends data to Grafana for monitoring device-specific download size, component breakdown, team/module/library contributions, and large files. For reduction, the post highlights optimization of DEX file size by addressing transitive R classes and refining resource management. Containment strategies are mentioned as long-term solutions.
  3. How We Cut GrabFood.com’s Page JavaScript Asset Sizes by 3xThis post details strategies for reducing JavaScript asset sizes on GrabFood.com by optimizing dependencies. It outlines a 7-step process: 1. Identifying large/duplicate dependencies using webpack-bundle-analyzer, 2. Investigating dependency usage via top-down and bottom-up approaches, and 3. Applying 7 specific techniques to reduce dependencies: lazy loading, unifying duplicate modules, using ES Modules for tree-shaking, replacing libraries with Web APIs, changing technical approaches to avoid large dependencies, avoiding Node.js dependencies, and optimizing external dependencies. The post quantifies the impact of these optimizations on various performance and cost metrics.
  4. Keeping 170 Libraries Up to Date on a Large Scale Android AppThis post details the process of keeping over 170 libraries up-to-date weekly for Grab's Android superapp. It outlines the rigorous assessment process for adding new libraries, the challenges of shared responsibility for updates, and a risk management framework based on library usage and the number of changes introduced. Strategies to reduce risk include frequent, incremental updates, investing in robust automation testing (UI tests on CI), and involving QA sign-off for high-risk updates. The post also emphasizes the importance of keeping libraries updated for compatibility with new Android API requirements.
  5. Returning 575 Terabytes of Storage Space to Our UsersThis post details the implementation of a cleanup mechanism to remove old, unused cache folders from the Grab app on user devices. This was identified as a significant contributor to app storage footprint. The mechanism automatically removes these old cache folders upon app update, resulting in the reclamation of up to 1GB of data per user and a total of 575 terabytes across over 13 million devices. The post also highlights the use of Android vitals and StorageManager API for instrumentation and data analysis to identify the root cause (discontinued third-party libraries leaving behind cache data) and measure the impact of the cleanup.