ByteDance Software Engineer mock interview

ByteDance HQ (Beijing/Singapore) — distinct from TikTok-US/UK. LeetCode Hard heavily overrepresented. HM round includes 15-min Hard DP problem. Beijing roles effectively require Mandarin. Pedigree-heavy: Tsinghua/PKU/Fudan dominate.

  • 3 rounds + HM
    Standard loop
  • LC Hard
    Coding bar
  • 15 min
    HM Hard DP limit
  • Free
    First mock

Why ByteDance HQ is structurally different from TikTok-US/UK

Per ByteDance HQ candidate reports on Glassdoor / Blind / 1Point3Acres + nodeflair's 243-interview compilation, ByteDance HQ (Beijing/Singapore/occasional US) is a meaningfully different filter from TikTok-US/UK despite same parent company. The bar is higher: LeetCode Hard heavily overrepresented vs TikTok-US's Medium emphasis. The HM round signature is a Hard DP problem with a 15-minute time limit — unusually short for the difficulty class.

Three signals matter most for screening: (1) Hard DSA fluency under time pressure — practice LC Hard with 15-30 min limits, not 45-60 min; (2) architecture-level depth on resume-listed tech — ByteDance probes anything you list (K8s scheduler internals, MySQL clustered index, Redis cluster sharding); (3) Mandarin / Chinese tier-1 alignment — Beijing/Shanghai roles effectively require Mandarin; international candidates need compensating signals (FAANG internships, Chinese tier-1 internships, target school pedigree).

Compensation by geography (per Levels.fyi May 2026): Beijing 1-2 ¥454K → 3-1 ¥1.71M; Singapore 1-2 S$115K → 3-1 S$326K; US 1-2 $198K → 2-2 $420K. ByteDance officially ended 996 in November 2021 (per Washington Post / IT Pro), replaced with '1075' — but Glassdoor reviews still describe high-pressure metric-driven culture. ByteStyles culture pillars (Always Day 1, Be Candid and Clear, Seek Truth and Be Pragmatic) get explicitly probed in behavioral rounds.

What the loop looks like

Round-by-round

  1. 01

    Recruiter screen

    Pre-onsite
    30 min

    Background, language fluency, location preference. Beijing/Shanghai roles effectively require Mandarin even when interview is offered in English. Singapore/global roles more flexible.

  2. 02

    HackerRank OA

    Pre-onsite
    ~90 min

    DSA-focused timed coding. LeetCode Medium-Hard mix. Auto-rejection below threshold. ByteDance OA bar is higher than TikTok-US OA — same parent company, different filter.

  3. 03

    R1 — Technical

    Onsite
    60 min

    LC Medium-Hard coding. Verbal CS fundamentals: MySQL clustered index, Redis data structures, TCP vs UDP, SSL/TLS, OS basics. Architecture-level depth on resume-listed tech.

  4. 04

    R2 — Technical

    Onsite
    60 min

    More coding (often LC Hard) + brief system design. DP, graph algorithms (DFS/BFS), binary search, heaps, tries, sliding window, segment trees.

  5. 05

    R3 — Hiring Manager

    Onsite
    60-90 min

    Project deep-dive + system design. Often includes a LeetCode Hard DP problem with 15-minute time limit. CAP-theorem reasoning, sharding, fault tolerance probed.

  6. 06

    HR / negotiation

    Post-onsite
    30 min

    Compensation discussion. Past R3, this is mainly negotiation, not screening. Cross-team 5th interview occasionally added — normal at ByteDance, not a red flag.

Question bank

Real interview questions reported by candidates

  • Coding

    "LeetCode Hard DP under 15-minute time limit (HM round signature). Reported examples: longest palindromic substring variants, edit distance with constraints."

    Source · vervecopilot.com ByteDance LeetCode questions, nodeflair 243-interview compilation
  • System design

    "System design: TikTok For-You recommendation feed at scale. Cover candidate generation, ranking, retrieval, A/B testing infrastructure, real-time user behavior pipeline."

    Source · designgurus.io ByteDance system design questions, systemdesignhandbook.com ByteDance guide
  • System design

    "System design: Toutiao news ranking ingestion at scale. Cover ingest pipeline, deduplication, ranking model serving, real-time vs batch trade-offs."

    Source · systemdesignhandbook.com ByteDance guide
  • CS fundamentals

    "Verbal: explain MySQL clustered index. When does it help vs hurt? How does it interact with secondary indexes?"

    Source · nodeflair.com ByteDance interview compilation
  • CS fundamentals

    "Verbal: Redis data structures — when to use sorted set vs hash vs list. Explain Redis cluster sharding strategy."

    Source · nodeflair.com ByteDance interview compilation
  • Coding

    "Coding: graph problem with DFS/BFS twist. Specific reported patterns: shortest-path-with-constraint, topological-sort-cycle-detection variants."

    Source · nodeflair.com, vervecopilot.com ByteDance
  • Behavioral

    "Behavioral mapped to ByteStyles — 'Always Day 1' (scrappiness), 'Be Candid and Clear' (disagreeing with manager using data). Concrete STAR examples expected."

    Source · joinbytedance.com/culture official ByteStyles
  • Behavioral

    "Why ByteDance specifically (vs TikTok, Tencent, Alibaba, Meituan)? Generic 'I want big tech in China' fails — show specific product engagement."

    Source · Glassdoor candidate reports, Blind ByteDance threads
Common signals to fix

What gets you rejected at this level

  • Incomplete coding under time pressure

    Per nodeflair 243-interview compilation: even one unfinished Hard problem in the 15-min HM round window often kills the loop. ByteDance's coding bar is harder than TikTok-US — practice LC Hard under strict time pressure before the loop, not LC Medium.

  • Shallow knowledge of resume-listed tech

    ByteDance probes architecture-level detail on anything you list. Putting Kubernetes on resume = expect 'explain K8s scheduler internals'. Don't list tech you can't defend at architecture level.

  • Single strong negative review overrides positive consensus

    Per Blind ByteDance threads: candidates report passing 4/5 rounds positively but getting cut due to one strong negative review. Every round matters — no 'fall through to next' buffer.

  • Mandarin gap for Beijing/Shanghai roles

    Per Blind threads ('if you're not Chinese don't apply to ByteDance'): Beijing/Shanghai roles effectively require Mandarin even when interview offered in English. Daily work and team collaboration is Mandarin. Singapore roles more flexible. Confirm language requirements with recruiter early.

  • Generic 'I want big tech' motivation

    ByteDance distinguishes from Tencent/Alibaba/Meituan. Generic answers fail. Show specific engagement: ByteDance product family beyond TikTok (Toutiao, Lark, Douyin, ByteDance Cloud), recent papers from ByteDance Seed (AI research arm), or specific engineering blog posts.

How Applr's AI mock interview tracks ByteDance's rubric

Applr's ByteDance HQ mock simulates the LC Hard time-pressure format — 15-min limits on Hard DP problems matching the HM round structure. The 'verbal CS fundamentals' filter is reproduced (MySQL clustered index, Redis cluster sharding, TCP/UDP) — generic 'I used MySQL' answers get flagged with rewrites that surface architecture-level depth.

System design rounds prompt ByteDance-specific scale (TikTok For-You, Toutiao ranking, video CDN). Behavioral rounds map answers to ByteStyles framework (Always Day 1 / Candid / Truth-seeking / Courageous) — generic 'I want big tech' framing gets rewrites referencing specific ByteDance products beyond TikTok.

FAQ

Is ByteDance the same as TikTok?

ByteDance is the parent company; TikTok is one of its products (operating outside China — Douyin is the China version). Hiring is split: ByteDance HQ (Beijing/Singapore/some US) for the parent company's products (Toutiao, Lark, Douyin, ByteDance Cloud). TikTok-US/UK has its own org with separate hiring. Interview bar differs — ByteDance HQ harder than TikTok-US per aggregated candidate reports.

Is Mandarin required for ByteDance?

For Beijing/Shanghai HQ roles: effectively yes per Blind threads and candidate reports. Even when interview is offered in English, daily work and team collaboration is Mandarin-default. Interviewers may switch to Mandarin mid-round. Singapore office is more flexible — English-primary with Mandarin as plus. Pure international background without Mandarin is the weakest profile for HQ roles. Confirm language expectations with recruiter early.

What's ByteDance salary in different geos?

Per Levels.fyi (May 2026): Beijing 1-2 (new grad) median ¥454K, 2-2 ¥786K, 3-1 ¥1.71M, top of band ¥2.53M. Singapore 1-2 ~S$115K, 2-1 ~S$189K, 3-1 ~S$326K. US 1-2 ~$198K, 2-1 ~$294K, 2-2 ~$420K. PPP-adjusted, Beijing comp at senior is competitive with US. Singapore sits in between.

What's the 'HM 15-min Hard DP' really like?

Per vervecopilot.com ByteDance LeetCode guide: at the hiring manager round, ByteDance regularly assigns one LeetCode Hard DP problem with a 15-minute time limit. This is unusually short for Hard DP (typical FAANG = 30-45 min for similar). Practice DP under aggressive time pressure: longest palindromic substring, edit distance variants, palindrome partitioning. Solving in 15 min on Hard DP separates serious applicants.

What about the Tsinghua / PKU pedigree thing?

Per 36kr reports: ByteDance is a top-2 destination for Tsinghua grads. Tsinghua has dedicated ByteDance recruitment channels. Heavy over-representation of Tsinghua/PKU/Fudan/SJTU among ByteDance HQ engineers. International candidates without these credentials need compensating signals: FAANG internships, top Chinese tier-1 internships (Tencent/Alibaba/Meituan), substantive open-source contributions. Pure international background without Mandarin AND without target pedigree is hardest.

Did ByteDance end 996?

Yes officially in November 2021 (per Washington Post / IT Pro confirmed). Replaced with '1075' (10am-7pm, 5 days). HOWEVER: Glassdoor / Indeed / Blind reviews still describe high-pressure metric-driven culture, daily metric reviews, team-dependent overtime. The 'wolf culture' (狼性) framing is broader Chinese tech context, not exclusive to ByteDance. Don't expect FAANG-level work-life balance even post-996.

Ready to practice for ByteDance?

Start free →