The Python Developer Projects That Actually Create Conversation

Me: Tell me about your weather app.

Candidate: It takes a city name and shows the temperature
using the OpenWeather API.

Me: What happens if the API is slow?

Candidate: It waits.

Me: Okay. Anything else in the project you would like
to talk about?

Candidate: No, that is the main feature.

Me: Tell me about your weather app.

Candidate: It started as a simple API call. Then I noticed
the free plan has a call limit, so I cached each city's
result for ten minutes. I also set a 3-second timeout,
because one slow response froze the whole page.

Me: Where is the cache stored?

Candidate: In memory with a dict and a timestamp first.
I moved it to Redis when I ran two worker processes,
because each worker had its own copy.

import time
import requests


def fetch_with_retry(url, attempts=3, timeout=5):
    """Fetch a URL, retrying on network errors and 5xx responses."""
    for attempt in range(1, attempts + 1):
        try:
            resp = requests.get(url, timeout=timeout)
            if resp.status_code < 500:
                resp.raise_for_status()
                return resp.json()
        except (requests.ConnectionError, requests.Timeout) as exc:
            if attempt == attempts:
                raise
            print(f"attempt {attempt} failed: {exc}")
        time.sleep(2 ** attempt)  # 2s, 4s, ... backoff
    raise RuntimeError(f"{url} kept returning server errors")

This small function gives me at least five things to ask:

1. Why retry only on 5xx and not on 4xx?
   (A 404 will not fix itself; a 503 might.)
2. Why exponential backoff instead of a fixed sleep?
3. Why does raise_for_status() sit inside the if?
   (HTTPError is not caught, so 4xx fails fast.)
4. Is it safe to retry a POST the same way?
5. How would you test this without hitting the real URL?
   (Mock requests.get with unittest.mock.patch.)

A candidate who wrote this can answer all five.
A candidate who copied it usually cannot answer #3.

import pytest
from quiz.scoring import score_answers


def test_all_correct():
    assert score_answers(["a", "b"], ["a", "b"]) == 2


def test_unanswered_counts_as_zero():
    assert score_answers(["a", None], ["a", "b"]) == 1


def test_answers_are_case_insensitive():
    assert score_answers(["A"], ["a"]) == 1


def test_length_mismatch_raises():
    with pytest.raises(ValueError):
        score_answers(["a"], ["a", "b"])

def score_answers(given, correct):
    """Return the number of correct answers.

    given   -- list of the user's answers (None if skipped)
    correct -- list of the right answers, same length
    """
    if len(given) != len(correct):
        raise ValueError("answer lists must be the same length")
    return sum(
        1
        for g, c in zip(given, correct)
        if g is not None and g.strip().lower() == c.strip().lower()
    )

PROJECT BRIEF

Name: ______________________________
One-line purpose (who uses it, for what): ________

Stack: Python ___ | Framework ___ | DB ___ | Other ___

The hardest problem I hit:
__________________________________________

The choice I made and the option I rejected:
__________________________________________

What I gave up because of that choice:
__________________________________________

How I know it works (tests, logs, numbers):
__________________________________________

ASK YOURSELF THESE 8 QUESTIONS OUT LOUD

1. Why Python for this, and why this framework?
2. Where is the data stored, and why there?
3. What happens when input is empty or malformed?
4. What happens when an external call fails?
5. What would break first if usage grew a lot?
6. Which part would you rewrite today?
7. How did you test the most important function?
8. What did you learn that a tutorial did not teach?

If any answer is 'I don't know', fix the project
or prepare an honest answer before the interview.

README STRUCTURE THAT HELPS THE INTERVIEWER

## What it does (2 lines)
## How to run it
    python -m venv .venv
    pip install -r requirements.txt
    python -m app
## Design decisions
    - Chose X over Y because ...
## Known limits
    - Does not handle ... yet
## Changelog
    - v0.2: added retry and timeout after slow API calls
## Tests
    pytest -q