Project Background

Meta's Facebook Android and iOS teams want to improve Feed performance in low-connectivity environments by revisiting what data is cached locally versus fetched on demand. You are the mobile engineer leading execution for a 12-week project spanning Feed ranking surfaces, media loading, and session persistence. Leadership wants a launch plan before the next quarterly planning review because Feed cold-start latency and data usage complaints have increased in emerging markets.

Key Stakeholders

The Feed Product Director wants faster app opens and better session continuity. The Infra lead is concerned about increased device storage, stale data, and cache invalidation complexity. Privacy and Legal require that sensitive user data and deleted content respect retention rules. The Ads team is worried that aggressive local caching could serve stale sponsored content or break impression logging.

Constraints

The project has 12 weeks, 6 engineers total (3 iOS/Android client engineers, 1 server engineer, 1 data engineer, 1 TPM), and no additional headcount. The team has a fixed budget of $180K for experimentation, QA, and monitoring improvements. The app must stay within an additional 120 MB max local storage budget per user, and launch cannot regress crash-free sessions below 99.85%. Any server API changes require approval from the Feed backend team by Week 4.

Complications

1. A parallel initiative to reduce app size is targeting a 15 MB binary reduction this quarter, creating tension with any new offline assets or cache libraries.
2. A recent policy update requires user-deleted posts and privacy setting changes to propagate to clients within 15 minutes.
3. The Ads team has already committed to a monetization experiment in Week 10, limiting launch windows.

Your Task

1. Define a decision framework for what Facebook Feed data should be cached locally versus fetched on demand.
2. Create a 12-week execution plan, including milestones, owners, and dependencies.
3. Propose a launch strategy, including experiment design, rollout stages, and rollback criteria.
4. Identify key risks around stale data, privacy, storage, and monetization, with mitigation plans.
5. Specify success metrics and guardrails for the first 30 days after launch.