Live Interview with a Junior Candidate: Technical Q&A and Feedback

Programmer Yupi hosts a live interview with a recent graduate, referred to as "A Qiang," who has half a year of internship experience, personal projects, and prior interviews at a major tech company.

The interview begins with a discussion of the candidate's experience at ByteDance, noting the intensity of a one‑hour interview with over 50 questions.

Yupi proceeds to ask a series of technical questions covering a wide range of backend topics:

1) Redis data structures and HyperLogLog drawbacks – the candidate discusses various structures and mentions project experience.

2) Precise PV/UV counting and supporting million‑level concurrency – the candidate proposes local atomic counters, periodic reporting, message queues, and log collection.

3) Understanding of AQS and classes built on it – the candidate correctly explains AQS concepts and cites ReentrantLock and CountDownLatch.

4) Concurrent programming classes and thread‑pool troubleshooting – mentions custom thread pools and CompletableFuture, admits limited production troubleshooting experience.

5) Handling billion‑scale data writes and data loss prevention – describes a pipeline involving Kafka, Flink, Hive, and S3.

6) Server‑side Gzip compression behavior – explains when servers return Gzip based on Accept‑Encoding.

7) Apollo configuration center features and release process – outlines usage and deployment workflow.

8) LiteFlow rule engine implementation – discusses configuration, persistence, orchestration, hot‑refresh, and context isolation.

9) Choosing Dubbo over Feign and implementing Dubbo – compares RPC frameworks, admits limited knowledge of custom implementation.

10) Pooling technology design – covers connection reuse, pool sizing, idle management, timeout handling, and thread safety.

11) Designing an extensible SDK – suggests Java SPI mechanism for plugin architecture.

12) Project deployment and automated pipeline creation – mentions Linux, Jenkins, but notes gaps in JVM tuning and observability.

13) Single Sign‑On implementation – explains token‑based SSO and mentions missing shared‑cookie approach.

14) Stress‑testing considerations

15) ThreadLocal usage cautions

16) Designing a custom caching annotation

After the technical round, Yupi offers the candidate a salary range of 200k‑300k RMB and provides interview advice, emphasizing the importance of showcasing live projects, adding more projects to the résumé, and structuring system‑design answers with a clear, top‑down approach.

Additional resources such as interview video recordings and related articles are linked throughout the post.