Meta Engineering Manager Interview Questions

1.Tell me about the last time you had to ship something quickly to hit a deadline, then iterate based on what you learned. What was the initial version, and how did you evolve it?

2.Describe a project where you had to iterate quickly after launch because the initial approach wasn't working. What did you change and how fast were you able to pivot?

3.Walk me through how you decide what your team should work on next quarter. What's your framework for prioritizing between competing projects?

4.Tell me about a time you had to give critical feedback to someone on your team about their performance or approach. How did you deliver it and what happened next?

5.Describe a time you championed a technical or product direction that your team or leadership initially pushed back on. How did you build support for it?

6.How do you think about the social impact of the products your team builds? Give me a specific example of how this influenced a technical or product decision.

7.Your team owns the graph serving layer that powers Facebook's People You May Know feature. The ML team wants to add a new signal that requires real-time friend-of-friend traversals, but this could impact query latency. How do you evaluate this request?

8.You're launching a new Threads feature next week, but your load testing shows the backend can only handle 60% of expected traffic. The product team says the launch date can't move. What do you do?

9.Your team maintains a service that handles Instagram story uploads. A new product requirement needs real-time story reactions, but adding this would increase database write load by 50x. Walk me through your technical approach.

10.A critical bug in WhatsApp message delivery is affecting 5% of users randomly. Your oncall engineer can't reproduce it locally, and the error logs aren't showing obvious patterns. How do you lead the incident response?

11.You're scaling the News Feed ranking system and need to add a new ML feature that requires storing 10TB of embeddings. Your team wants to build it fast using existing infrastructure, but the ML team says we need a specialized vector database. How do you break this tie?

12.You need to migrate a legacy Python service that processes Facebook notifications to a new Rust-based architecture. The service handles 100M notifications daily and can't have downtime. What's your migration strategy?

13.Design the real-time video call system for WhatsApp, handling 50 million concurrent calls globally. How do you ensure low latency and handle network partition scenarios?

14.Design a feature flagging system that every single request across Meta's apps would hit, serving flags for A/B tests and gradual rollouts. The system needs 99.99% availability even when its primary datastore is completely down.

15.Design a content moderation system for Instagram Reels that can process 1 billion uploads per day. The system needs to detect harmful content within seconds while minimizing false positives.

16.Your team needs to build a cross-posting feature that lets creators publish the same content to Facebook, Instagram, and Threads simultaneously. How do you design this while respecting each platform's unique content formats and audiences?

17.Design the friend suggestion algorithm infrastructure that powers 'People You May Know' across Facebook's 3 billion users. The system needs to generate suggestions in real-time while being privacy-conscious.

18.Design the infrastructure for Meta Quest's multiplayer VR experiences, supporting up to 100 users in a single virtual space with real-time avatar movement and voice chat. How do you handle the unique constraints of VR latency requirements?

19.Tell me about a time you had to convince a peer engineering team to adopt a technical approach that would benefit your team but create more work for theirs. How did you approach it?

20.Describe a situation where you had to make a decision quickly with incomplete information that affected multiple teams. What was your process?

21.You're leading an engineering team that's been tasked with improving Instagram's creator monetization features, but your team has never worked on payments or commerce systems. How do you approach this challenge?

22.Walk me through how you've coached an engineer who was technically strong but struggled with product sense or understanding user impact. What specific techniques did you use?

23.Your team is responsible for a core infrastructure service that other teams depend on, and you need to deprecate a widely-used API to improve system performance. Three major product teams are pushing back hard. How do you navigate this?

24.Tell me about a time you had to completely change your team's roadmap mid-quarter because of a strategic shift or new priority. How did you maintain team morale and momentum?

25.Instagram Reels is seeing 15% higher engagement in India compared to the US, but our revenue per user is 80% lower. How would you estimate the long-term business impact if we optimized the algorithm specifically for Indian users?

26.Your team's API that serves Facebook's news feed ranking is experiencing 2% higher latency during peak hours. Nothing changed in your code, but this impacts 2 billion users. Walk me through how you'd diagnose this systematically.

27.WhatsApp wants to add message reactions, but each message currently gets 50M views on average in busy group chats. How would you estimate the infrastructure cost of storing and serving reaction data?

28.Meta's family of apps shares user identity across Facebook, Instagram, and WhatsApp. If we wanted to build a unified notification center that aggregates notifications from all three platforms, how would you estimate the technical complexity and infrastructure requirements?

29.Instagram's Explore feed algorithm needs to be retrained daily using fresh user interaction data from 2 billion users. If each user generates 500 interactions per day and each training run costs $50K in compute, how would you optimize this to reduce costs while maintaining recommendation quality?

30.Meta Quest's multiplayer VR experiences need to maintain sub-20ms latency for 50 users in the same virtual space. If each user generates 1KB of positional data per frame at 90fps, and you have datacenters in 15 regions globally, how would you architect the real-time synchronization system?