← all repositories
ShanghaiTechAIClub/DLInterview

A cram sheet for deep-learning job interviews, in Chinese

Curated question lists for ML/DL roles, plus a mirror of CyC2018's broader CS interview notes.

1.1k stars Learning
DLInterview
Velocity · 7d
+0.4
★ / day
Trend
steady
star history

What it does

This repo collects interview questions across six categories—math, machine-learning fundamentals, deep learning, computer vision, algorithms, and programming languages—aimed at candidates for Chinese deep-learning positions. It also repackages CyC2018’s well-known 2018 CS interview notes (networking, OS, Java, distributed systems, etc.) as a bundled reference.

The interesting bit

The value is in the curation, not invention: the DL questions are crowd-sourced and still being gathered, while the bulk of the repo is a mirror of another project’s book-derived study notes. Think of it as a shared flash-card deck with a table of contents.

Key highlights

  • Six topical markdown files for DL-specific prep (Math, ML, DL, CV, Algorithms, Language)
  • Includes CyC2018’s extensive notes on OS, networking, databases, Java, distributed systems, and coding practice
  • Sourced from standard texts (e.g., Computer Networking 7th ed., Modern Operating Systems, Introduction to Algorithms)
  • Notes are manually typed and accept PRs for corrections or additions
  • Automation tooling mentioned for Markdown conversion and image handling (GFM-Converter, Text-Typesetting)

Caveats

  • The DL question collection is explicitly “still gathering” (正在收集题目中); depth and coverage are unclear without opening each markdown file
  • No English translation provided; entirely Chinese-language content
  • Last major activity appears tied to 2018-era material; freshness of DL-specific questions is uncertain

Verdict

Worth bookmarking if you’re interviewing for Chinese ML/DL roles and read Mandarin. Skip it if you need English material, hands-on coding problems, or up-to-date LLM-era coverage.

heatdrop uses Google Analytics to see which pages get read — nothing else. Your call. How we handle data.