What Python engineering practices do you consistently apply?

Good Python code is boring in the best sense: obvious control flow, clear names and few surprises. PEP 8 and a formatter such as Ruff handle the mechanical style so reviews can focus on correctness, design and operability.

from dataclasses import dataclass

@dataclass(frozen=True)
class RetryPolicy:
    attempts: int = 3
    base_delay: float = 0.2

    def delay_for(self, attempt: int) -> float:
        return self.base_delay * (2 ** attempt)

def fetch_with_retry(client, url, policy=RetryPolicy()):
    for attempt in range(policy.attempts):
        try:
            return client.get(url)
        except TimeoutError:
            if attempt == policy.attempts - 1:
                raise
        time.sleep(policy.delay_for(attempt))

The non-negotiables

• Never use mutable default arguments or mutable class attributes for per-instance state
• Use context managers for anything that must be closed or released
• Add type hints on public interfaces and run a checker in CI
• Automate formatting and linting with Ruff; automate tests with pytest
• Pin application dependencies and commit a lock file

Prefer explicit over implicit and simple over clever. A function that fits on one screen and does one thing is easier to test and review. When a piece of code needs a comment to explain what it does, first try renaming or restructuring it.