LinkedIn DevOps / SRE Interview Questions

1.Tell me about a recent time when you gave or received difficult feedback about your infrastructure work that significantly changed how you approached a problem. Walk me through what happened.

2.Walk me through a time when you took ownership of an infrastructure problem that was clearly outside your team's direct scope. Why did you step in and how did you drive it to resolution?

3.Tell me about a time you pushed back on an infrastructure change or feature request because you believed it would negatively impact member experience. What was the situation and how did you handle it?

4.Describe a situation where you raised the engineering bar for your team - maybe around monitoring, deployment practices, or infrastructure standards. What was broken and how did you drive the change?

5.Describe a time when you had to build a working relationship with a difficult product engineering team or stakeholder to deliver a critical infrastructure project. How did you approach it?

6.Tell me about a calculated infrastructure risk you took that didn't work out as expected. What was your reasoning going in, what went wrong, and what did you learn?

7.LinkedIn's canary deployment process automatically promotes changes from 1% to 100% traffic if metrics look good. What key metrics would you monitor during a canary for a service that handles member profile updates?

8.You notice LinkedIn's Java microservices are experiencing increased garbage collection pauses during peak traffic. The services use LinkedIn's JVM auto-tuning, but GC times are still impacting P99 latency. How do you investigate and resolve this?

9.LinkedIn deploys changes to production within 30 minutes via trunk-based development. A critical microservice starts failing intermittently after your deployment, but only affecting 2% of traffic. Walk me through your incident response approach.

10.Write a Python script that monitors Kubernetes pod restart rates across our microservices and alerts when any service exceeds normal restart patterns. Focus on the alerting logic.

11.You're designing monitoring for LinkedIn's new machine learning inference service that powers job recommendations. It needs to handle 100k QPS with strict P99 latency SLAs. What observability strategy would you implement?

12.LinkedIn's feed ranking service uses Kafka for real-time updates and batch jobs for model training. How would you design the infrastructure to handle both streaming and batch workloads efficiently?

13.LinkedIn Learning serves millions of video streams globally. Design the content delivery and caching infrastructure to ensure consistent playback quality for learners worldwide, especially in regions with limited bandwidth.

14.LinkedIn's microservices architecture has 3000+ services communicating via Rest.li and gRPC. Design a service mesh observability platform that helps engineers debug cross-service issues and understand the blast radius of service failures in real-time.

15.Sales Navigator processes millions of search queries daily from sales professionals. Design the search infrastructure to handle peak usage while maintaining sub-200ms response times, considering that search patterns are highly skewed toward popular companies and titles.

16.LinkedIn Jobs recommendation engine needs to process member activity data in real-time to update job suggestions. Design the data pipeline that can handle 50k events per second while ensuring recommendations reflect the latest member interactions within 5 minutes.

17.LinkedIn Feed serves billions of personalized posts daily. Design the infrastructure to pre-compute and serve feed rankings while handling the thundering herd problem when a viral post from an influencer with millions of followers gets published.

18.LinkedIn Recruiter handles sensitive candidate data across different regions with varying privacy regulations. Design a data storage and access control system that ensures compliance while allowing recruiters to efficiently search and manage candidate pipelines globally.

19.Tell me about a time you had to influence engineers across multiple product teams to adopt a new infrastructure standard or tool. How did you build consensus?

20.LinkedIn's observability team is asking all service owners to adopt their new distributed tracing solution. Several teams on your platform are resistant because it requires code changes. How do you drive adoption?

21.LinkedIn's engineering culture emphasizes fast iteration and continuous deployment. Describe a time when you had to slow down or block a deployment to maintain system reliability. How did you handle the pushback?

22.A junior engineer on your team deployed a configuration change that caused a service outage affecting Sales Navigator users. They're feeling terrible about it. How do you handle the situation with the engineer and with stakeholders?

23.You discover that LinkedIn's Kafka infrastructure is experiencing message lag during peak hours, potentially affecting feed updates for millions of members. Multiple product teams are asking for updates. How do you lead the incident response?

24.You need to migrate LinkedIn's legacy monitoring system to a new platform, but it will require temporary reduced visibility during the transition. Three different VP-level stakeholders have conflicting priorities about timing. How do you build alignment?

25.LinkedIn's member profiles are updated 200 million times per day globally. Estimate the infrastructure cost impact if we reduced our profile cache TTL from 1 hour to 15 minutes.

26.You notice that LinkedIn Jobs search latency spikes every Monday morning at 9 AM PST but not on other days. Walk me through how you'd investigate this pattern.

27.LinkedIn Learning serves 25 million monthly learners globally. Estimate the storage and bandwidth costs if we increased video quality from 720p to 1080p for all courses.

28.Sales Navigator's prospect search needs to handle 10x traffic during a major industry conference week. Current capacity is 50k searches per hour. Design your scaling approach.

29.LinkedIn's notification system processes 500M daily notifications across email, mobile, and web. Estimate the cost impact if iOS notification delivery rates dropped from 95% to 85% due to Apple policy changes.

30.LinkedIn's Feed ranking model serves 15B feed impressions daily. The ML inference service is experiencing 99.5% availability but we need 99.9%. Estimate the infrastructure investment required.