This guide draws on 8 first-hand interview accounts from Nowcoder. Candidates describe about 1 round on average, most often technical interviews and behavioral interviews.
These 8 writeups cover data, software engineering roles. All of them were posted under a forum handle, so we could not verify the authors’ identities.
New grad. 1 round. Shared by AIGC小白入门记 on nowcoder, 1970-01-01
1First roundTechnicalAgent infrastructure questions on isolation, recovery, sandbox scaling and context compression.Covered: Biggest engineering challenge, Multi-tenant isolation, Task state recovery, Sandbox pool scheduling, Evaluating generated code, Goal drift in long tasks, Security-focused review agent, RAG vs long context, Context compression
Java Developer
Not selected
New grad. 1 round. Shared by 西一昂 on nowcoder, 2026-09-14
1Second roundTechnicalHalf an hour on the internship, a few Java and Redis questions, an easy coding task and career outlook questions.Covered: Internship deep dive, Agent project, Redisson reentrant locks, Java IO models, Connections vs threads, Merge sorted arrays
Backend Engineer
Outcome not shared
unknown. 1 round. Shared by 无名小兵 on nowcoder, 1970-01-01
1Second roundTechnicalA second round dominated by conceptual questions about agents, MCP and skills.Covered: Agent use cases in projects, Using agents to fix production issues, MCP, LLM vs agent, Harness concept, Skill formats and structure, Multi-agent collaboration, Evaluating agent output
AI Infrastructure Engineer
Outcome not shared
New grad. 1 round. Shared by Code_Agent on nowcoder, 1970-01-01
1InterviewTechnicalProject grilling followed by in-depth Kubernetes control-plane and GPU batch scheduling questions, plus Ray internals.Covered: Idempotent Kubernetes controllers, Informer list-watch and caching, resourceVersion semantics, Watch recovery and etcd compaction, API Priority and Fairness, Scheduler assume and bind, Kueue admission and cohort borrowing, Volcano gang scheduling, Preemption and starvation, GPU fragmentation, Device plugin restarts, Ray scheduling and object store, Idempotent retries
Java Developer
Outcome not shared
New grad, Beijing. 1 round. Shared by upc on nowcoder, 2026-09-15
1Second roundTechnical50 minutes split into fundamentals, project discussion and a sliding-window coding problem.Covered: Design patterns, Thread-safe lazy initialization, Java memory model, Ordered multi-thread printing, Spring circular dependencies, Agent project and MCP, Longest substring without repeats
Embedded Software Engineer
Not selected
New grad. 1 round. Shared by zeertzjq on nowcoder, 2026-09-16
1Second roundBehavioral40-minute round with light resume questions and mostly behavioral prompts.Covered: Resume questions, Project difficulties, Failure experience, Handling pressure, Strengths and weaknesses, Career plans
Test development engineer
Outcome not shared
New grad. 1 round. Shared by 懒羊羊豁达 on nowcoder, 2026-09-24
1Second interviewBehavioralRelaxed, conversational round on projects, habits and motivation.Covered: Project understanding, Python scripts and common commands, AI-assisted productivity, Stress relief, Impressions of Xiaomi, Three-to-five-year plan
AI frontend intern
Outcome not shared
Internship. 1 round. Shared by 牛客51638671号 on nowcoder, 2026-09-25
1First interviewTechnicalAn hour of coding, frontend fundamentals and open-ended AI engineering scenarios. Likely passed.Covered: Two coding problems on shared screen, Debounce and throttle, HTTP, WebSocket and SSE, git merge vs rebase and reverting, Live performance profiling in DevTools, Vue rendering and React architecture, Forward vs reverse proxy, Backpressure for fast LLM token streams, Team-wide AI coding knowledge base, Quality control for AI-written code and E2E testing
This guide draws on 8 first-hand interview accounts from Nowcoder. Candidates describe about 1 round on average, most often technical interviews and behavioral interviews.
What topics does Xiaomi test in interviews?
Commonly reported topics include Biggest engineering challenge, Multi-tenant isolation, Task state recovery, Sandbox pool scheduling, Evaluating generated code, Goal drift in long tasks.
These guides summarize public, first-hand interview experiences shared by candidates. They describe the shape of each loop and the topics that came up, not a leaked question bank, and every experience links back to its original source. Processes change often. Treat this as directional prep, not a script.
TheJobsMap