← Back to Directory

ByteDance interview notes, page 13 of 29

FAANG / Big Tech

Most of these ByteDance write-ups come from Software Engineers (194), then Machine Learning Engineers (45) and Data Scientists (17). The most reported rounds are the technical phone screen (227), the system design round (28), the behavioral / leadership round (16) and the onsite coding round (9).

289 interview notes · updated Sep 2026

ByteDance·Software Engineer·Technical Phone Screen

May 2026
Bytedance coding round, one algorithmic problem about linked lists. Pretty niche problem type, not your typical easy-medium grind.
  • Given the head of a linked list, reverse the nodes within each group of even length. The list is split into groups of sizes 1, 2, 3, 4, and so on. Reverse a group if its size is even, leave it alone if odd. The last group might be smaller than expected, but the rule still applies based on its actual size.

“I'd seen linked list reversal before but never with this grouping twist.”

View Post

ByteDance·Software Engineer·Technical Phone Screen

May 2026
SRE interview at Bytedance, technical phone screen focused entirely on a single deep-dive scenario around web service degradation. No fluff, just one long question that kept branching. Left feeling like I covered maybe 70% of what they wanted.
  • You get paged that a web service is slow. Walk through your full triage process from the moment you see the alert to when you've mitigated the issue.
  • Which Linux tools would you reach for at each layer of the investigation, and what specifically are you looking for in their output?
  • Once you've gathered data from your tooling, how do you isolate which component is actually responsible for the slowness?

“This is the kind of question that sounds manageable until you realize they want the whole stack.”

View Post

ByteDance·Software Engineer·Technical Phone Screen

May 2026
Bytedance SWE interview, got a cache design problem that looked manageable on the surface but had enough edge cases to keep me busy for the whole session.
  • Design an in-memory key-value cache with LRU eviction and per-entry expiration. Each entry has a time-to-live, expired entries should be lazily removed when accessed, and get/put should both run in O(1) average time.

“I knew the LRU part cold, hashmap plus doubly linked list, done that a dozen times.”

View Post

ByteDance·Machine Learning Engineer·Technical Phone Screen

May 2026
Bytedance ML engineer interview that went deep on Transformer internals fast. They had me write pseudocode live while fielding theory questions at the same time, which was more disorienting than I expected.
  • Write pseudocode for scaled dot-product self-attention.
  • Write pseudocode for cross-entropy loss.
  • What is the role of the position-wise feed-forward network inside a Transformer block?
  • Why are attention scores scaled by 1 divided by the square root of the key dimension?

“I got through the mechanics fine: compute Q, K, V projections, dot Q with K-transpose, scale, softmax, multiply by V.”

View Post

ByteDance·Software Engineer·Technical Phone Screen

May 2026
Bytedance software engineer interview with a word search problem that had a pretty nasty optimization twist. The core question was straightforward but the constraint about 30,000 dictionary words made it a real trie problem, not just a DFS.
  • Given an m x n grid of letters and a large list of dictionary words, find all words from the list that can be formed by tracing a path through adjacent cells (up, down, left, right) without reusing any cell. Optimize for a very large dictionary rather than searching independently for each word.

“My first instinct was to just do DFS for each word separately.”

View Post

ByteDance·Software Engineer·Onsite - Coding / Algorithms

May 2026
ByteDance SRE onsite with a LeetCode problem at the very end. Tight on time but managed to pass all 15 test cases on the first run, which felt pretty good.
  • Solve a coding algorithm problem under time pressure, with test cases run live immediately after you finish writing.

“Only had about 10 minutes left when they dropped the coding problem on me.”

View Post

ByteDance·Software Engineer·Technical Phone Screen

May 2026
Bytedance SWE interview, got a classic grouping problem that I've seen a dozen times but still managed to second-guess myself on the implementation details.
  • Given an array of strings, group all the anagrams together and return the groups in any order.

“I knew this one.”

View Post

ByteDance·Software Engineer·Technical Phone Screen

May 2026
Bytedance software engineer interview with at least one coding problem that looked like LeetCode 620 or something close to it. Not a ton of detail to go on but the problem itself is worth knowing.
  • Given a table of movies, write a SQL query to find all movies that have at least one description that is not 'boring', and return them ordered by rating descending. (LeetCode 620 or similar variant)

“Pretty standard SQL filter and sort, nothing tricky about the logic itself.”

View Post

ByteDance·Software Engineer·Technical Phone Screen

May 2026
Went through a technical screen for a Research Scientist role at Bytedance and got hit with a classic sliding window problem. Nothing too surprising but the pressure of the setting made me second-guess an approach I've done a dozen times before.
  • Given a string, find the length of the longest substring that contains no repeated characters.

“I knew this was a sliding window problem the moment I read it, but I still fumbled the edge cases around spaces and symbols.”

View Post

ByteDance·Software Engineer·Onsite - System Design / Architecture

May 2026
System design round at Bytedance for a software engineer role. The whole session was basically one big auction system question with a lot of back-and-forth on architecture choices, which was actually kind of fun given I work with this stack daily.
  • Design a real-time auction system that can handle high read and write volumes, especially during peak bidding periods.

“I initially said WebSocket for everything and had to walk it back mid-discussion.”

View Post