This guide draws on 10 first-hand interview accounts from Nowcoder. Candidates describe about 1 round on average, most often technical interviews, a hiring manager interview and behavioral interviews.
These 10 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 roundTechnicalRL training questions followed by designing an LLM agent for ride-hailing scenarios.Covered: RL framework and agent loop, GRPO, On-policy vs off-policy, Customer service dialogue state, Real-time ride requests, Tool orchestration for booking, Peak-time fallbacks, Agent quality metrics
Algorithm engineer
Outcome not shared
New grad. 1 round. Shared by 那位谢大人 on nowcoder, 2026-09-20
1Technical interviewsTechnicalMostly internship discussion plus RL and LLM fundamentals. Second round grilled an agent project.Covered: Internship deep-dive, PPO/GRPO/DAPO, Agentic RL data, Reward hacking, Transformer implementation, Agent memory
Backend developer intern (efficiency platform)
Not selected
Internship. 1 round. Shared by 鹏游蝶梦 on nowcoder, 2026-09-20
1First technical interviewTechnicalInternship project walkthrough followed by MySQL and RAG design questions.Covered: Internship project design, MySQL slow query and EXPLAIN, Covering index, Optimistic locking, RAG multi-path retrieval, BM25 and Milvus, RAG quality assurance
AI Infra Algorithm Engineer
Outcome not shared
unknown. 2 rounds. Shared by 算法打工人小旺 on nowcoder, 2026-10-01
1First roundTechnicalDeep questions on LLM inference and training internals, then CUDA kernels for normalization and softmax variants.Covered: Attention optimization, KV cache management, MLA and linear attention, GQA memory estimates, Pipeline scheduling, DeepSpeed ZeRO, AdamW memory, CUDA kernels
2Second roundTechnicalProbed the candidate's scheduler work and PyTorch compiler internals, with an LRU cache coding task.Covered: Scheduling optimization, Distributed decode, PyTorch dispatch, torch.compile internals, CUDA Graph, LRU cache, Chunked prefill, Triton debugging
Backend Engineer (Mobility)
Outcome not shared
New grad. 1 round. Shared by 辣条吃多了_ on nowcoder, 2026-09-16
1Third roundHiring managerSenior interviewer guided an open-ended discussion of internship work, improvements, AI strategy and motivation.Covered: Internship challenges, Project shortcomings, Cross-team collaboration, AI productivity, Customer-facing AI ideas, Recent learning, Motivation
Backend engineer
Outcome not shared
New grad. 1 round. Shared by Harry00 on nowcoder, 2026-09-23
1First interviewTechnicalFifty-five minutes on project data models, permissions and indexing, ending with a sliding-window problem.Covered: Project deep-dive, RBAC model and schema, Redis data structures, MySQL indexes and B+ trees, Composite index usage, Count of length-K substrings without repeats
Backend engineer
Outcome not shared
New grad. 1 round. Shared by 辣条吃多了_ on nowcoder, 2026-09-26
1HR interview (after three technical rounds)BehavioralFinal HR conversation on internships, motivation and career plans, with offer timeline discussed.Covered: Internship comparison and takeaways, Self-study, Why still job hunting despite a return offer, Other companies interviewed, Knowledge of the team, Short, mid and long-term plans
Role not stated
Outcome not shared
unknown. 1 round. Shared by kong123456 on nowcoder, 1970-01-01
1Written testOnline assessmentMultiple-choice fundamentals plus two coding problems. The candidate fully solved the first and partly the second.Covered: 20 multiple-choice fundamentals questions, Minimum cooling operations (binary search), Pick k flowers maximizing value plus distinct-count squared
Role not stated
Outcome not shared
unknown. 3 rounds. Shared by aacgll on nowcoder, 2026-09-12
1First roundTechnicalInternship-focused first round.Covered: Internship engine design, protocols and caching, Writing a prompt, A few fundamentals
2Second roundTechnicalProject and design questions plus a backtracking coding task.Covered: Internship project, Order state machine design, Backtracking problem
3Third roundTechnicalAgent design and AI workflow questions. All three rounds took place on the same day.Covered: AI collaboration and learning, Daily development workflow, Designing and evaluating an agent, Building a knowledge base, Vector index structures
Backend Developer (Mobility)
Outcome not shared
New grad. 1 round. Shared by 辣条吃多了_ on nowcoder, 2026-09-12
1Second roundHiring managerA leader round looking for business and technical judgment beyond implementation details.Covered: Internship deep dive: edge cases and performance, Production incidents and lessons, Big-picture thinking about the business, AI productivity work: spec coding and knowledge base, Evaluation and memory systems, Skills and harness engineering
What Didi is hiring for now
10 open roles in our index. Advertised pay floors run $149k to $170k (median $170k, from 8 postings that state pay).
This guide draws on 10 first-hand interview accounts from Nowcoder. Candidates describe about 1 round on average, most often technical interviews, a hiring manager interview and behavioral interviews.
What topics does Didi test in interviews?
Commonly reported topics include RL framework and agent loop, GRPO, On-policy vs off-policy, Customer service dialogue state, Real-time ride requests, Tool orchestration for booking.
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