This guide draws on 22 first-hand interview accounts from Nowcoder. Candidates describe about 1 round on average, most often technical interviews.
These 22 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. 2 rounds. Shared by 球球给个offer子 on nowcoder, 1970-01-01
1First roundTechnicalA friendly 40-minute round of big-data fundamentals with one algorithm and one SQL problem.Covered: Project architecture choices, MapReduce, Hive vs Spark, Complex SQL, Alert handling, Data skew, Warehouse layering, Spark job, stage and task, Data quality for live-stream payments, Maximum subarray sum, Consecutive-streak SQL
2Second roundTechnicalAbout 50 minutes, half spent grilling project details from ingestion to modeling, plus Spark questions and a SQL task.Covered: End-to-end project deep dive, Warehouse layering theory, Data skew, Spark tuning and execution flow, Wide vs narrow dependencies, Kappa architecture, Window-function SQL
C++ Developer
Outcome not shared
unknown. 1 round. Shared by 服务端老cpp on nowcoder, 1970-01-01
1Second roundTechnicalC++ language-design questions on parameter passing, numeric conversions and type erasure.Covered: Parameter passing by value vs reference vs rvalue, Floating-point conversion pitfalls, Void pointers vs type erasure, Memory allocation
1Second roundTechnicalProject-based database and agent questions, then building a log-analysis tool with an AI coding assistant.Covered: Schema design, Sharding, Isolation levels, MVCC and phantom reads, Inventory deduction, RAG evaluation, Agent memory, AI-assisted coding
AI Full-Stack Engineer
Outcome not shared
New grad. 1 round. Shared by 明天会有好天气 on nowcoder, 2026-09-14
1First roundTechnicalInternship chat then a broad round of Java, MySQL and Redis internals ending with an island-area coding problem.Covered: MySQL insert path and two-phase commit, Garbage collection and ZGC, Slow SQL, AI coding workflow, Repeatable read, Redis locks and data structures, Kafka rebalancing, Max area of island
AI full-stack engineer
Not selected
New grad. 1 round. Shared by Gwen0x4c3 on nowcoder, 2026-09-21
1Second interviewTechnicalBackend fundamentals followed by an AI coding task to scope, build and test a small product.Covered: Spring @Transactional pitfalls, MySQL isolation and phantom reads, Redis big keys, MySQL replication, AI-assisted product build
Java backend engineer
Outcome not shared
New grad. 1 round. Shared by YukiYukino on nowcoder, 2026-09-21
1Second interviewTechnicalA ninety-minute round that drilled each Java and storage topic to its limits. The candidate passed.Covered: LRU cache coding, Spring bean lifecycle and AOP, Thread pool rejection policies, HashMap thread safety, volatile, Garbage collection, Coroutines, MySQL B+ tree indexes, Redis AOF
Frontend engineer
Outcome not shared
unknown. 1 round. Shared by patience_1 on nowcoder, 2026-09-21
1First interviewTechnicalProject design questions followed by an AI-assisted coding exercise.Covered: DFS cycle handling, pgvector, BM25 and RRF choices, Store sharding, Parse and render buffering, Incomplete SSE lines, AI coding task
Frontend engineer intern (e-commerce)
Outcome not shared
Internship. 1 round. Shared by patience_1 on nowcoder, 2026-09-21
1Second interviewTechnicalReact and SSE fundamentals with no project questions, plus a custom hook coding task.Covered: React and one-way data flow, useMemo vs React.memo, useEffect vs useLayoutEffect, SSE and legacy browser fallback, localStorage-synced state hook coding
Test Development Engineer (Payments)
Outcome not shared
unknown. 1 round. Shared by 三本咋了 on nowcoder, 2026-09-30
1First roundTechnicalOne-hour round on fundamentals and internship agent work, ending with an AI-coding exercise instead of hand-written code.Covered: Java collections, Concurrency, Processes vs threads, Thread pools, MySQL indexes, Redis persistence, LLM agent tool routing, AI-assisted coding
AI Application Developer
Outcome not shared
New grad. 1 round. Shared by Code_Agent on nowcoder, 1970-01-01
1First roundTechnicalProject questions on an AI compliance-review agent followed by Java, JVM and MySQL fundamentals.Covered: Project architecture, Prompt design, Layered memory for multi-round review, Read vs write tools, Verifying model findings, Multi-route retrieval and HyDE, Why similar retrieval can still be wrong, HashMap internals and thread safety, JVM memory areas, Stack overflow, Off-heap leaks, GC algorithms and ZGC, B+ tree indexes, Optimizer choosing full scans
AI Application Developer
Outcome not shared
New grad. 1 round. Shared by Code_Agent on nowcoder, 1970-01-01
1Second roundTechnicalDeeper agent-engineering round on durable memory, structured generation, multi-system consistency, fine-tuning tradeoffs and retrieval operations.Covered: Resuming long workflows after crashes, Concurrent memory updates with versioning, Small model for memory retrieval, Compacting JSONL memory logs, Constrained decoding for structured output, Incremental JSON validation, Compensation across external systems, SFT vs DPO data, Safety vs accuracy after DPO, RAG evaluation with incomplete labels, Lost-in-the-middle, Embedding model migration, HNSW deletions
Java Developer (Kling platform)
Outcome not shared
New grad. 1 round. Shared by 同尘去 on nowcoder, 1970-01-01
1First roundTechnicalA friendly but deep first round that probed internship design decisions, flash-sale and sharding scenarios, and ended with coding an LRU cache.Covered: N+1 calls and pooled parallel aggregation, Thread pool sizing per downstream, Timeouts in aggregation, Scaling aggregation 1000x, Production JVM memory triage, MDC propagation and OpenTelemetry, Evaluating a keyword intent filter, Spring event threading, Coupon flash sale without oversell, Redis Lua stock deduction failures, Resharding from 8 to 16 tables, RocketMQ idempotency, Index design and failure, LRU cache
AI Application Backend Intern
Outcome not shared
Internship. 1 round. Shared by 可乐roue on nowcoder, 2026-09-16
1First roundTechnicalDeep questions on a risk-control agent project, a backend platform project, Redis and RocketMQ, then an easy coding problem.Covered: Multi-agent topology, Context passing and explosion, Agent memory and evaluation, Post-launch tuning, RAG chunking and embeddings, Redis skew, big and hot keys, RocketMQ duplicates and backlog, Two sum
Frontend Engineer Intern (Live E-commerce)
Not selected
Internship. 1 round. Shared by 风止何安啊 on nowcoder, 2026-09-17
1First roundTechnicalInternship deep dive on AI chat frontends and SSE, React basics, then an AI-assisted coding task explained via its README.Covered: Learning with AI, Frontend-to-LLM pipeline, Multi-agent task splitting, Frontend performance, SSE reconnection and stalled streams, Multi-agent visualization, React optimization and keys, Deep vs shallow copy, AI coding: API replay and response diff
Test Development Engineer Intern (Local Services)
Received an offer
Internship. 1 round. Shared by 批踢踢 on nowcoder, 2026-09-17
1Second roundTechnicalClassic test design exercises, internship discussion and a min-stack coding problem. Verbal offer followed.Covered: Understanding of test development, Triangle program test design, int-to-float boundaries, Internship, Prompting for code quality, Min stack
Test Development Engineer Intern (Local Services)
Received an offer
Internship, Beijing. 1 round. Shared by 批踢踢 on nowcoder, 2026-09-17
1First roundTechnicalJava and middleware fundamentals, internship quality questions, a SQL task and AI coding workflow. Passed.Covered: Motivation for switching, Java reflection and collections, Redis structures and hot keys, Message queues, Content correctness and evaluation, SQL: manager of highest-paid department, AI coding context management
Backend developer
Outcome not shared
New grad. 1 round. Shared by 艾坤日记 on nowcoder, 2026-09-28
1First interviewTechnicalInternship and agent questions plus backend fundamentals, finishing with one coding problem.Covered: Single versus multi-agent design, MCP versus local CLI tools, Agent observability metrics, Human versus LLM as judge, Code knowledge base design, KV cache reuse, Protobuf versus JSON, QPS optimization and measurement, Kafka and zero copy, Minimum window substring
Data Engineer
Outcome not shared
New grad. 2 rounds. Shared by 球球给个offer子 on nowcoder, 1970-01-01
1First roundTechnicalA friendly 40-minute round of big-data fundamentals with one algorithm and one SQL problem.Covered: Project architecture and storage choices, MapReduce workflow, Hive vs Spark, Complex SQL from resume, Alert types and handling, Data skew diagnosis, Warehouse layering, DWS vs DWM layers, Spark driver, job, stage and task, Data quality checks for live-stream revenue, Maximum subarray sum, SQL: longest consecutive streak
2Second roundTechnicalAbout 30 minutes of detailed project grilling followed by easier Spark and warehouse questions and a SQL task.Covered: End-to-end project deep dive, Warehouse layering theory, Data skew, Spark tuning and execution flow, Wide vs narrow dependencies, Personal view of data warehousing, Kappa architecture, SQL: windowed averages
Backend Developer Intern (Feed)
Outcome not shared
Internship. 1 round. Shared by Fu1222 on nowcoder, 1970-01-01
1First roundTechnicalA relaxed 50-minute round on internship work and AI coding with one easy linked-list problem and no trivia.Covered: Internship project deep dive, Understanding and use of AI coding, Merge two sorted lists (ACM I/O mode)
Data Engineer
Outcome not shared
New grad. 2 rounds. Shared by 球球给个offer子 on nowcoder, 1970-01-01
1First roundTechnicalA friendly 40-minute round of big-data fundamentals with one algorithm and one SQL problem.Covered: Project architecture and storage choices, MapReduce workflow, Hive vs Spark, Complex SQL from resume, Alert types and handling, Data skew diagnosis, Warehouse layering, DWS vs DWM layers, Spark driver, job, stage and task, Data quality checks for live-stream revenue, Maximum subarray sum, SQL: longest consecutive streak
2Second roundTechnicalAbout 30 minutes of detailed project grilling followed by easier Spark and warehouse questions and a SQL task.Covered: End-to-end project deep dive, Warehouse layering theory, Data skew, Spark tuning and execution flow, Wide vs narrow dependencies, Personal view of data warehousing, Kappa architecture, SQL: windowed averages
LLM Application Developer
Outcome not shared
New grad. 1 round. Shared by AIGC小白入门记 on nowcoder, 1970-01-01
1Second roundTechnicalProject questions followed by detailed reinforcement learning math and RAG design.Covered: Agent project end-to-end flow, Target users and scenario of a dialogue AI project, Whether Q&A informs or makes decisions, Agentic RL vs traditional RL, Instability in long-horizon agent RL, PPO clipping vs KL constraint, GRPO loss vs PPO, GAE formula, lambda and gamma, Importance sampling and high-variance weights, RAG pipeline and chunking strategy
Role not stated
Not selected
unknown. 1 round. Shared by 咏之吟 on nowcoder, 2026-09-18
1First roundTechnicalInternship-only discussion with no fundamentals, coding or AI coding task. Rejected the same evening despite positive feedback.Covered: Agent platform knowledge base, AI usage details in internship
This guide draws on 22 first-hand interview accounts from Nowcoder. Candidates describe about 1 round on average, most often technical interviews.
What topics does Kuaishou test in interviews?
Commonly reported topics include Data skew, Hive vs Spark, Warehouse layering, Maximum subarray sum, End-to-end project deep dive, Warehouse layering theory.
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