Learn GenAI
Github
Contents
Map

Quiz · 01 · Python & Systems

20 questions from 4 pages

These are the Check Yourself questions from each page of the module, collected in course order. Each heading links back to the page the questions test. All module quizzes →

Python for AI Engineering

Check yourself
0 / 5 answered
  1. Why commit uv.lock for an application?
  2. You run 5,000 model calls with asyncio.gather and get mostly HTTP 429 errors. What's the first fix?
  3. Why add random jitter to retry backoff?
  4. Which work should go to a process pool rather than asyncio or threads in standard CPython?
  5. Why test LLM application code with fake model responses rather than real calls?

API Design for LLM Services

Check yourself
0 / 5 answered
  1. A document-analysis endpoint takes 2-6 minutes per request. Which interaction style fits best?
  2. Users report the chat answer appears all at once after 15 seconds, although the backend streams. What is the most likely cause?
  3. A client times out waiting for a POST that starts an agent run, and retries. How do you avoid running the agent twice?
  4. Which status code and header should a saturated LLM API return when it sheds load?
  5. Why should a streaming endpoint check for client disconnects?

Git Workflows for ML

Check yourself
0 / 5 answered
  1. Where should a 40 GB training dataset used by several experiments be versioned?
  2. An API key was pushed to a public repository an hour ago. What is the first thing to do?
  3. Why should prompt changes go through pull requests with an eval gate?
  4. Quality dropped somewhere in the last 64 commits. Roughly how many eval runs does git bisect need to find the culprit?
  5. Why strip notebook outputs before committing?

Linux & the GPU Box

Check yourself
0 / 5 answered
  1. nvidia-smi shows 'CUDA Version: 13.0'. What does that tell you?
  2. Your training job was killed by a spot-instance reclaim and lost six hours of progress. What change prevents that?
  3. PyTorch DataLoader workers crash with 'bus error' inside a Docker container. What is the likely fix?
  4. How do you call a vLLM server running on port 8000 of a remote GPU machine from your laptop without opening the port to the internet?
  5. dmesg shows repeated 'Xid 79' for one GPU. What does it mean, and what do you do?