LLMs Interview Notes

What LLMs Interview Notes does

The repository collects more than 800 frequently asked interview points for AIGC, large model and AI agent roles. Each question follows the same structure: a one-sentence answer first, then the reasoning, then follow-up questions an interviewer might ask. It is organized by topic rather than by article or company, and each question carries a difficulty rating.

The module list in the README includes RAG with 160 questions, agents with 100, reinforcement learning post-training with 120, GraphRAG with 100, Loop Engineering with 100 and object detection with 100. A Hermes Agent module is listed with 100 planned questions, of which 22 have detailed walkthroughs and answer points so far, linking to a hermes_agent.md document. The README is written in Chinese and invites issues for new questions and pull requests for corrected answers. It is aimed at candidates changing from computer vision, NLP or backend work, and at interviewers who need questions. The license is Apache-2.0.

Key features

• 800+ interview points across LLM and agent topics
• Hermes Agent module with 100 listed questions
• One-sentence answer, reasoning and follow-ups per question
• Difficulty ratings on questions
• Open to issues and pull requests

When to use it

• Preparing for an AI agent interview that asks about Hermes Agent
• Writing interview questions as a hiring engineer
• Reviewing RAG, agent and post-training topics in Chinese

Who it is for: Chinese-reading candidates and interviewers for large-model and AI agent engineering roles.

How it fits with Hermes Agent

Hermes Agent is one topic module among many, so this is study material rather than Hermes tooling.

Note: The content is written in Chinese and the Hermes Agent module is only partly completed.

FAQ